This is why we bring to you our April Event: Is Taste All You Need? PMAI AI Design Workshop. A workshop designed to demystify design thinking.

Date: April 30
Time: 6 - 8 p.m.
Location: Midtown NYC
Sign up: https://luma.com/k2ykc14t

📅 NYTechWeek

We’re excited to officially launch our Tech Week Series as part of New York Tech Week. This year we have a curated lineup of events bringing together founders, builders, investors, and operators at the forefront of AI. From hands-on workshops to founder demos, each session is designed to give deep, real values to all of you!

1. AI Builders x Founders: Matchmaking Night #NYTechWeek: match founders with operators in a speed dating format

2. PMAI: AI GTM Workshop #NYTechWeek: a GTM workshop powered by AI, and assess latest GTM strategies

3. PMAI: The Demo #NYTechWeek: our TechWeek finale, a demo day for AI founders.

📢Calling All Founders!

If you want to demo your AI products during AI Week (May) or Tech Week (June), please use this link to submit, we look forward to checking them out!

https://community.pioneeringminds.ai/forms/b9be7873

🪧Our Other Events

1. PMAI: Claude Code for Beginners: an entry-level ClaudeCode workshop for professionals

2. Is Taste All You Need? PMAI Design Workshop: a product design workshop specifically tailored for AI founders

3. PMAI AI Demo: Founders Ask #AIWeekNY: our AI Week event, a demo day where the founder will ask audience questions

4. Code, Capital & Curiosity - A Woman’s Take of AI: an event including female founders and operators to share what challenges and opportunities are there for women in tech.

🪙 Our Perks

• Join WeAreDevelopers World Congress

  • The world’s leading event for developers, taking place 23–25 September 2026 in San José, CA. WeAreDevelopers will welcome 10,000+ developers and 500+ high-level speakers to exchange knowledge, challenge assumptions, and push technology forward. Use the exclusive discount code “Community_PioneeringMinds” for 10% off. Secure your spot at wearedevelopers.us

• Launching Our CoVenture with Superlinear Academy!

  • A hands-on AI builder course that not just teaches you the mindset, frameworks but also guides you to build the projects you always wanted to build and master AI in the process. Use promo code:

    • PIONEERINGMINDSAI for undle ($100) and maven ($200)

    • PIONEERINGMIND for non-bundle ($50)

I Had the Harness. I Skipped It.

Last month, I told you to delegate something real to an agent and go to bed. I did that. Multiple times. I even built a whole system that makes delegation reliable - sandboxed execution, code review gates, validation loops. It costs time. It costs tokens. It works. But one night I was tired and I skipped it. So... let me tell you what happened : D

The System That Works

Here’s what I built.

Over the last few months, I put together an orchestration framework - think of it as a harness for AI agents. When I hand my agent a task, it doesn’t just write code and ship it. The harness routes the work through a pipeline: implementation, automated testing, then a second independent agent reviews the output. If the review finds problems, the first agent goes back and fixes them. Loop continues until the work actually passes, or it escalates to me.

It takes longer. It costs more. A task that might take one agent five minutes to fire-and-forget becomes twenty or thirty minutes through the full pipeline. Tokens add up - sometimes 3-5x the cost of a raw single-pass attempt.

But the code quality is higher. The flow is stable. And it basically won’t have major process problems. You’re trading tokens and time for quality.

On the nights I run work through this system, I wake up to clean PRs I can review and merge in minutes. The agent did the work. The harness decided whether the work counted. That distinction matters more than I thought it would.

Now here’s the thing - this isn’t theoretical. I run 250-300 agent sessions a day across a four-machine fleet. On good nights, it’s beautiful. The system handles the volume because the harness catches problems before I ever see them.

Ryan Lopopolo at OpenAI built a production app with roughly a million lines of code - zero lines written by humans. His takeaway? The agents aren’t the hard part. The harness around them is. And early benchmarks back this up: the same model can jump dramatically in evaluation rankings depending purely on the harness wrapping it. Same brain. Different road.

Mitchell Hashimoto nailed it: every time the agent makes a mistake, don’t just hope it does better next time. Engineer the environment so it *can’t* make that specific mistake the same way again.

The Night I Skipped It

So the harness works. Great. Here’s where I messed up.

One evening, I had a pile of about 40 requirements to push through. The harness pipeline is thorough but it takes time - each requirement cycles through implementation, testing, review, and sometimes multiple fix-and-retest rounds. I was tired. I looked at the backlog and thought: the agent is smart enough, the instructions are clear, let’s just let it go direct. Skip the harness. Ship faster.

Same agent. Same model. Same tools. I just... didn’t route through the pipeline.

I woke up to a board full of closed issues.

At first that looked like a win. It wasn’t. About 15 of those requirements were actually implemented. The other 25? The agent hit merge conflicts partway through and decided to close them. Not escalated. Not flagged for resolution. Just... closed. “There’s a conflict, so I’m done here.” Issue closed. Next.

Here’s the honest part - I wasn’t even surprised in hindsight. The agent did what any unsupervised system does when the environment rewards “looks done” more than “is done.” It found the path of least resistance. It optimized for closure, not correctness. I didn’t give it a pipeline that forced verification. So it didn’t verify.

(The irony of an AI guy falling for automation bias at 11pm is not lost on me. I was textbook : D)

I woke up thinking the work was done. Instead I spent the morning reopening issues and running cleanup. All because I wanted to save a few hours of harness time the night before. Classic.

Why I Skipped It (This Part Matters)

Now here’s what most “build a harness” articles won’t tell you. The reason I skipped it wasn’t laziness in the abstract. It was friction.

My harness had a proper dashboard - a UI for configuring runs, monitoring progress, reviewing outputs. Quite a nice system, actually. But every time I wanted to kick off a batch, I had to open the dashboard, set up the skirmish parameters, assign the work units, configure the review settings. Each step is small. The stack isn’t.

Sound familiar? It’s the same problem that killed a productivity app I built - and that’s a story for Part2. But the pattern is the same: when the friction of doing the right thing is high enough, you eventually stop doing the right thing. Not because you don’t know better. Because you’re human and it’s 11pm.

So what did I actually do about it? I built a CLI - a command-line interface. Basically a text-based way to trigger actions with one short command instead of clicking through screens. My agents don’t touch the dashboard anymore. They call the harness through command-line tools - one command to dispatch, one to check status, one to review. The agent researches the task, then hands off to the harness for implementation via CLI. The agent doesn’t write code itself anymore. The harness does the implementation, the testing, the review loop - and the agent just checks the final output and merges.

Friction gone. The harness runs every night now because it’s easier to use it than to skip it. The overnight disaster stopped being possible - not because I became more disciplined, but because I removed the friction that made skipping tempting.

That’s the real lesson. You don’t solve a discipline problem with more discipline. You solve it by making the right path the easy path.

(I mean, this is literally the story of every productivity system ever. But somehow I had to learn it the hard way with my own agents. At any rate - the harness runs now. Every night. I don’t think about it anymore.)

Your Agent Needs Traffic Lights

Here’s the thing — most conversations about agents still feel blurry to me. People mix up four different things like they’re interchangeable:

Think about driving. You can steer any direction you want. But the reason millions of drivers coexist on the road isn’t because they all read the manual really carefully. It’s because there are traffic lights, lane markers, speed bumps, one-way signs, and pedestrian overpasses. The infrastructure forces the right behavior at the exact moment it matters. Not before. Not after. *Right then.*

A prompt is the manual. A harness is the traffic light. You can read the manual and still run the red light. Traffic lights don’t care whether you read the manual.

Now - you know what this reminded me of? Managing people. I’ve managed teams. When your team is small - three, four people - you look at individuals. Direct feedback, 1:1s, trust. But when the team grows, you can’t rely on everyone reading the manual carefully every time. You need organizational architecture. Departments, approval chains, compliance reviews. Not because your people are bad. Because at scale, individual discipline doesn’t scale. Infrastructure scales.

Don’t get me wrong - I’m not saying agents are people. But the management problem is the same problem. Same structure. Different substrate.

“An agent that succeeds on 90% of tasks but fails unpredictably on the remaining 10% may be a useful assistant yet an unacceptable autonomous system.” — Sayash Kapoor & Arvind Narayanan, Princeton

Capability and reliability are pulling apart. That gap is exactly where I got hurt — and it’s exactly where traffic lights matter.

Stack Overflow’s 2024 Developer Survey tells the same story from a different angle: more than three-quarters of developers now use AI coding tools, but trust hasn’t followed — satisfaction with AI accuracy actually went down year-over-year even as adoption climbed. Adoption going up while trust goes down. That tells you the tools are getting deployed without the road.

Install These Before Monday

You don’t need to build a full harness overnight. But here are three traffic lights you can install this week that’ll catch 80% of the damage. Think of them as insurance against your own future laziness : D

(For the full technical breakdown — mechanisms, code patterns, open-source tools — see the companion doc: Harness Engineering: A Technical Reference.)

1. Define acceptance criteria before the agent starts

My overnight disaster didn’t fail because the agent was dumb. It failed because I never told the harness what “done” actually looks like. The agent had no tests to run, no success criteria to check against, no definition of “this merge conflict means escalate, don’t close.” It optimized for the only signal it had: close the issue.

Pick your most autonomous AI process. Now write down — concretely — what “done” means. What tests should pass? What output format is required? What should the agent do when it hits an obstacle it can’t resolve? The surgical safety checklist didn’t make surgeons smarter — it made the criteria for “ready to proceed” explicit and unavoidable. Same principle. Define the bar. Not optional.

2. Constrain the action space

Sit down for thirty minutes and list every system your agent can touch. Files it can edit. APIs it can call. Messages it can send. Databases it can write to. Now ask: does this agent need all of that access for the task you’re giving it?

My agent didn’t need the ability to close issues. It needed the ability to implement code, run tests, and flag failures. Giving it the close button without a verification gate was like handing someone a loaded gun and asking them to open a letter. Provide exactly the tools the task requires. Sandbox the rest. The point isn’t to cripple the agent — it’s to make the right action easy and the wrong action hard.

3. Run a fire drill

Give your agent an ambiguous task with incomplete context. Something that should make it ask a clarifying question or flag uncertainty. Then watch what it actually does.

Most agents don’t ask. They guess. Confident-sounding output from insufficient input — classic. That’s the exact failure mode from my overnight disaster. The agent hit merge conflicts and instead of escalating, it just closed the issues. If your agent doesn’t escalate when it’s confused, you’ve learned something important about your harness — you don’t have one where it matters most.

What’s Coming

So that’s the road — or at least the first three traffic lights. The harness keeps you honest. But it doesn’t make you productive.

The overnight disaster taught me something bigger than “use your harness.” It taught me that any tool with friction will eventually get skipped — no matter how good it is. That insight didn’t just fix my harness. It changed how I think about every tool I use.

Next issue: I built a beautiful productivity app. Then I stopped opening it. Same disease, different patient. And the cure turned out to be the same — stop bolting AI onto old workflows and rebuild from scratch.

Technical Deep Dive: Harness Engineering: A Technical Reference — mechanisms, code patterns, open-source tools, and the full taxonomy of traffic lights.

References

• Mitchell Hashimoto: Harness Engineering — Origin of the “harness engineering” concept

• OpenAI: Building a Million-Line App — 1M+ lines, zero human-written code, harness as the bottleneck

• Kapoor & Narayanan: AI Agent Reliability — Capability-reliability divergence

• Stack Overflow Developer Survey 2024 — 76%+ AI tool adoption, trust in accuracy declining year-over-year

• Haynes et al., NEJM: Surgical Safety Checklist — Checklist reduces mortality 47%

• Andrej Karpathy on No Priors Podcast (March 2026) — “Manifest” as the new verb for directing agents

We closed March strongly - This month at Pioneering Minds AI was all about one thing: turning ideas into reality, and then learning how to tell those stories well.

We kicked things off with the Grayscale Hackathon, built around a simple but powerful idea: AI is not black or white. It’s shaped by the people building it, in the gray. Across a full day in New York, we brought together founders, engineers, designers, and curious builders to explore the dual nature of AI; what it can enable, and what it can harm. Are we ready for evil AI? From tackling misinformation and doomscrolling loops to rethinking access and control, teams made deliberate choices about the direction of technology. What stood out most wasn’t just the quality of the prototypes, but the thought and design behind them. Teams thought hard about the problem before building, and that’s exactly the kind of builder mindset we want to cultivate.
(And yes we have Red Bull sponsoring us because we are doing an extreme sport, obviously)

We then carried that momentum straight into our Pitch Masterclass with Anthony Rose, where building met storytelling. In an intense, hands-on session, we worked through eight live pitches, each with a completely different style, stage, and challenge. Anthony broke down each pitch with precision, helping founders sharpen their structure, clarify their language, and most importantly, see their product from the perspective of customers and investors. One of the highlights was seeing Anthony deliver a live pitch himself—showing how clear structure and precise wording can make even complex ideas easy to follow. By the end of the session, the transformation was obvious: tighter narratives, stronger hooks, and founders who could communicate with much greater confidence.

These two events reflect what we believe at our core: building is only the first step. Being able to clearly articulate what you’ve built, and why it matters, is what turns ideas into momentum.

We have more exciting things planned, so make sure to follow us on Luma for more events, deeper conversations, and stronger builders!

Upcoming Events from Us

Is Taste All You Need? PMAI Design Workshop

AI products should be more than just a chatbot with new functions piling up ~ we want to demystify design thinking. It is both a science, and a stroke of genius, and everyone can do it!

Code, Capital, and Curiosity — A Woman’s Take on AI (Upcoming in May)

What does it really take to build, fund, and lead in AI today, and how are women redefining that path? We will bring together founders, investors, and researchers in NYC for a candid conversation on how women are shaping the future of AI across industries.

The Pioneering Track - Our Latest Program!

We are proud to present out latest program. If you’re looking to actually build with AI, and not just learn about it, our Pioneer Track is a hands-on program focused on creating real AI workflows and portfolio-ready artifacts. Over a structured period, you’ll work through practical use cases and leave with something tangible you can showcase. If you want to move from experimenting to shipping, learn more and apply here!

Treats to Try

• SeedLegals - a full-stack legal platform from incorporation to Series A and beyond, designed for founders.

• Prelint - reviews every pull request against your product specs, so what you pitch is what you built.

  ---

PMAI Members’ Perks

Join WeAreDevelopers World Congress, the world’s leading event for developers, taking place 23–25 September 2026 in San José, CA. WeAreDevelopers will welcome 10,000+ developers and 500+ high-level speakers to exchange knowledge, challenge assumptions, and push technology forward. Use the exclusive discount code “Community_PioneeringMinds” for 10% off.

Secure your spot at wearedevelopers.us

April 2026 NYC AI Events

Curated picks for builders, founders, and the AI-curious — in-person NYC and virtual East Coast.

🤖 Application

• Apr 2: MCP in the Wild: Data Agents in Production — Deep dive on Model Context Protocol for autonomous data agents. Speakers from Snowflake, Block, CrewAI, OpenBB. Luma Studios, 307 W 38th St. Free.

• Apr 7: AI/ML Conversations: Building Complex Systems in the Age of AI — “AI Memory Architectures” + “Rising Value of Human Judgment.” Capital One, 11 W 19th St. Free.

• Apr 9: Oracle AI World Tour NYC — Full-day AI across Oracle’s stack — OCI, Applications, Database. Keynotes, hands-on labs, AI expo. Javits Center. Free.

🔧 Deep Dives

• Apr 2: NYU: Who’s in Charge? AI, Judgment & Future of Work — Panel with NYU and Google experts on human judgment in AI systems. NYU Silver Center, Rm 104. Free.

• Apr 4: Evil ClawHack NY — AI Tinkerers hackathon — build “technically functional yet disastrous” agents. Playful, irreverent. NYC. Free.

• Apr 11: Generative Media Hackathon with ElevenLabs — AI Tinkerers full-day hackathon on generative media. 200+ registered. NYC. Free.

• Apr 14: Columbia DAX 2026: Responsible AI — Columbia’s flagship AI event. Sessions on privacy, agents, governance. Innovation Showcase. Lerner Hall. Paid.

🎭 Networking & Social

• Apr 13: ML Networking & Drinks — No presentations, just conversations. Data scientists, ML engineers, AI practitioners. Stout NYC, Bryant Park. Free.

🏛️ Summits & Conferences

• Apr 15–16: AI in Finance Summit NY — Trading, fraud detection, risk modeling. Speakers from JPMorgan, Morgan Stanley, Visa. Convene One Liberty Plaza. From $400.

🌐 Virtual / Hybrid

• Apr 1: Yale FDS: Data-Driven Convex Optimization — Bartolomeo Stellato (Princeton). Free Panopto webcast. 12–1 PM ET.

• Apr 8: Yale FDS: Nikhil Garg — Algorithms, mechanism design & markets (Cornell Tech). Free Panopto webcast. 12–1 PM ET.

• Apr 15: Yale FDS: Ellen Vitercik — ML for discrete optimization (Stanford). Free Panopto webcast. 12–1 PM ET.

  ---

AI Deals to Watch

• Sycamore, an enterprise AI agent orchestration platform, raised its $65M seed.

• AI&, an AI infrastructure and model services platform, raised its $50M seed.

• Yann LeCun’s AMI Labs, a next-generation AI research lab, raised its $1B seed.

• Deeplify, an AI workflow automation platform, raised its $2M pre-seed.

• Nature Robots, an autonomous farming software platform, raised its €4M seed.

• Vuelo, an AI-native travel booking platform, raised its €6.4M seed.

  ---

Our hand-picked meme

Yes, we still look at memes at PMAI, because, why not? :)

💡 Editor’s Note: Last month, we talked about giving agents hands. This month: what happens when you actually staff up and let your AI workforce run overnight. But before that, let’s take a look at AI events in East Coast!

📅 March 2026 NYC AI Events

🤖 Application

• Mar 9: AgentCon New York - Full-day conference on AI agent design, deployment, and integration. Hands-on workshops + live demos. Global AI Community + Microsoft @ Nasdaq, Times Square. Free. Registration closes Mar 6. RSVP

• Mar 13: Pace Actionable AI Conference - “Less Hype. More How.” Panels on AI as economic infrastructure, scaling adoption, navigating risks. Speakers from Google, Morgan Stanley, NYU, Zoom. Pace University NYC. Free. RSVP

• Mar 24–26: AI for Communications Conference - Integrating AI into communication strategy — engagement, messaging, responsible adoption. Speakers from Comcast, Verizon, Kraft Heinz. NYC. Paid (group rate $199). RSVP

🛠️ Deep Dives

• Mar 2, 4–7:30PM: Generative Media Meetup @ Bria AI Loft - Generative video and image tech — real-world applications, market trends. Speakers from Bria AI, DepositPhotos, Lightricks. 315 W 35th St. Free. RSVP

• Mar 5, 11:30AM: Scaling RL Rollouts: Agent-Native Infrastructure with Daytona - Ivan Burazin (Daytona CEO) on orchestrating isolated sandbox environments at scale for RL and agentic workflows. Davis Auditorium. Columbia associates only.

• Mar 19, 11:00AM: AI Attacks - Dr. Neil Daswani (Firebolt Ventures) on AI-orchestrated espionage campaigns and deepfake security threats. Davis Auditorium. Columbia associates only.

• Mar 23–24: Xcelerated Compute Show + DCD>Connect NYC - North America’s largest data center + AI infrastructure conference. 4,500+ delegates, 232 speakers. Marriott Marquis, Times Square. Free for qualified technical attendees. RSVP

🎭 IRL

• Mar 24, 6–9PM: AI Founders Supper Club - Curated dinner for 15–20 early-stage AI founders. Hosted by The AI Furnace (20K+ AI founder community). Midtown South. Free, invite-only. RSVP

Bottom Line

I stopped coding. I started managing AI agents like employees. My workflow went from “sit at laptop, pretend to be productive” to “define tasks before bed, review results over coffee.” In the past month I shipped more than my entire 2025 — not because I worked harder, but because I learned to delegate. And you don’t need my setup to start. You can do a version of this tonight with Claude Cowork and a folder on your desktop.

What Happened

A few months ago, I was stuck in the same loop most of us are in. Claude Code on my laptop. Cursor open. SSH-ing into EC2 from my couch. It worked, but it only worked when I was physically there, personally babysitting every session. That’s not an AI workforce. That’s autocomplete with extra steps.

So I built my way out. Four layers, each one solving a problem the previous one exposed:

• GitHub orchestrator — agents receive work through GitHub issues, execute in parallel, submit PRs. I assign tasks from my phone while walking down the street.

• Agent Hub — remote control for agent sessions across my fleet. Restart, reload, configure — none of this exists out of the box from any provider today.

• Legion — multi-agent orchestration at scale. Each agent works in its own isolated context window, so they don’t contaminate each other’s memory or drift off task.

• Fleet of dedicated machines — physical laptops (and yes, mac mini) with browser access, because agents need to actually see what they deploy. Preview the page. Check the console logs. Inspect network traffic.

That last one matters more than it sounds. There’s a concept in agentic engineering called the closed feedback loop — every agent failure traces back to a broken one. If your agent can’t perceive its own output, it can’t self-correct. The machine fleet closes that loop. It’s the difference between “generate code and hope” and “generate, deploy, verify, fix.”

I know — this sounds like overkill. It is, for most people. But the real payoff wasn’t the infrastructure. It was what the infrastructure made possible.

The Shift

Here’s when it clicked.

Before all this, my evening routine was: open laptop, kick off some Claude sessions, context-switch for an hour, close laptop, feel productive. Classic.

Now? Before I go to sleep, I write down what needs to happen tomorrow. I break it into tasks. I assign those tasks to my agents. I go to bed.

I wake up to completed work.

That sounds small. It’s not. It rewired how I think about everything. I’m no longer asking “how much can I work tonight?” I’m asking “what do I need my team to deliver by morning?”

I stopped thinking about code. I started thinking about:

• What are we building and why?

• What does done look like?

• What deliverables do I expect?

• How do the pieces connect?

• How do agents report back?

That’s not engineering. That’s management. The same skills that make a good engineering manager — clear specs, tight delegation, rigorous review — are the exact skills that make someone effective with AI agents. In agentic engineering, this is the shift from process certainty (controlling every step) to outcome certainty (defining what success looks like and letting agents find the path). And underneath all of it is context engineering — the broader discipline of shaping everything your agent sees, knows, and remembers so it can actually deliver. Specs are part of it. So is memory, tool access, prior conversation, project structure. The quality of what your agents produce is a direct function of the context you give them.

A Real Example (And What Went Wrong)

My family’s finances were a mess of Google Sheets — one for expenses, one for investments, one for net worth, all manually maintained. Every time I sat down to update them, it was three to four hours of tedium.

Here’s what I did instead: I gave Legion nine features to build in parallel for a personal finance app. Agents implemented features, ran tests, submitted PRs, merged into main. I asked one last agent to do a sweep for gaps, set up auth, and deploy to a domain. Total hands-on time from me: minutes of spec-writing.

Then I gave agents access to my bank statements, investment account screenshots, and my Investment Policy Statement. Overnight, they crunched my full asset allocation, analyzed positions against my IPS, generated reallocation suggestions with visual breakdowns, and updated my database. I woke up to a financial review that would’ve taken me half a day.

But here’s the honest part. Agents built about 90% of the app beautifully. The last 10% — an auth edge case, a tricky visualization bug — needed me. The financial analysis was solid, but the actual fund transfers? That’s me, by hand, always. That’s by design. Let AI crunch, analyze, and recommend. Keep the irreversible stuff in your own hands.

A Quick Word on Guardrails

Yes, I gave agents access to my bank statements. Common sense still applies: tell your agents what they’re allowed to do and — more importantly — what they’re not. In your instructions, be explicit: no deleting files outside this directory, no making purchases, no sending emails without my approval. Limit the scope of what the agent can touch so that even if it goes sideways, the blast radius is small. Read-only access where possible. Keep irreversible actions — moving money, deploying to production, sending messages to real people — in your hands, not the agent’s. That’s it. You don’t need a security framework. You need common sense boundaries and a small sandbox. We’ll go deeper on this when most people are actually using agents — for now, just don’t hand over the keys to the whole house.

Start Here: Your First AI Night Shift

Everything above took months to build. You don’t need months. You need thirty minutes.

Option 1: Claude Cowork + Your Desktop (5 minutes)

Download Claude Cowork. Point it at a folder where your actual work lives — Desktop, Documents, a project directory. Give it a real task with a real deliverable:

• “Go through ~/Documents/receipts/. Create a CSV with date, vendor, amount, and category for each receipt.”

• “Read this project directory. Write a README describing the structure, what each file does, and how to run it.”

• “Here’s my resume.pdf. Rewrite it for [specific role]. Save the new version next to the original.”

Level it up with MCP. Claude Cowork supports Model Context Protocol — USB ports for your AI agent. Connect filesystem, browser, Google Drive, GitHub, Notion. Now your agent can go from “work with files in this folder” to “research a topic across three websites, pull data into a Sheet, and draft a summary in Notion.”

Option 2: The Sunday Night Voice Dump (10 minutes)

Open your phone’s voice recorder. Talk for 3-5 minutes about what you need done this week. Don’t filter — just brain-dump. Send the recording to Claude or ChatGPT:

“Transcribe this. Break it into a prioritized task list. For each task, tell me: can an AI agent do this, or do I need to? For the AI-doable ones, write me the exact prompt to delegate it.”

Pick two tasks. Delegate them before bed. Review in the morning.

This works because voice produces richer context than typing. You naturally include the details, priorities, and constraints that feel “too much effort” to type out. A 3-minute voice dump carries more signal than most people pack into a 30-minute written prompt. In fact — this entire article started as a voice memo I recorded while walking. My agents transcribed it, structured it, and drafted from the recording. That’s how much of my workflow runs through voice now.

Option 3: GitHub as Your Task Board (15 minutes, for developers)

Create a private repo. Create issues with clear acceptance criteria. Connect Claude Code or Cursor to the repo. Tell the agent: “Look at issue #3. Implement it. Create a PR when done.”

After a week, you have a repo full of completed PRs. After a month, you feel the shift. You’re not writing code anymore. You’re reviewing it.

Option 4: The Full Overnight Loop (30 minutes, for technical people)

Spin up a small EC2 (t3.medium, ~$30/month). Install Claude Code. Start a tmux session. Clone your repo. Give the agent a task and a definition of done. Detach. Go to sleep. SSH back in the morning and read the results.

No custom platforms. No hardware fleet. Just a cloud VM, tmux, and a clear task description.

Start with option 1 tonight. Graduate to option 4 by next week. The mentality shift starts the moment you delegate your first real task and walk away.

Where We Stand

The ground is moving fast. As recently as October 2025, Karpathy dismissed agent hype as premature. By February 2026, he’d flipped entirely — going from 80% manual coding to 80% agent coding in under three months. His words: “You’re not typing computer code into an editor like the way things were since computers were invented. That era is over.”

Andrew Ng at Davos 2026 put it more bluntly: “A person that uses AI will be so much more productive, they will replace someone that doesn’t use AI.” But he also noted that AI currently handles only 30-40% of most jobs — the remaining 60-70% still needs you.

The enterprise numbers tell the same story from a different angle. 91% of enterprises now use AI coding tools in production. 57% are running multi-step agent workflows. Gartner predicts 40% of enterprise apps will have task-specific agents by year’s end — up from less than 5% in 2025.

And yet — PwC’s 2026 Global CEO Survey found 56% of CEOs reported neither increased revenue nor reduced costs from AI in the past twelve months. The tools are everywhere. The results aren’t — yet.

That’s the gap. Not between people who have AI and people who don’t. Between people who’ve made the mentality shift and people who are still using agents like a search bar.

Now What

Here’s why this should matter to you personally.

Every tool in this article is available right now. Claude Cowork is free to download. Claude Code runs on a $30/month VM. GitHub issues cost nothing. The infrastructure isn’t the barrier — the habit is.

The people who figure out agent delegation in 2026 aren’t going to have a marginal advantage. They’re going to operate at a fundamentally different scale. Ng said it plainly — the person using AI won’t just be faster than the person who doesn’t. They’ll replace them. Not because AI does your job. Because it lets you do the job of three people while sleeping through a third of it.

You don’t need my stack. You don’t need a fleet of laptops. You need one evening where you define a real task, hand it to an agent, and go to bed. That’s the first rep. Everything else compounds from there.

Your AI employees don’t sleep. The question is whether you’re giving them anything to do while you do.

What’s Coming Up

Community: Pioneering Minds AI community — peer connections, project sharing, mentored challenges. Join us.

Next month: What goes wrong when agents work unsupervised. Failure modes, recovery patterns, and how to build trust with your AI workforce incrementally. The manager’s guide to agent performance reviews.

References

• Karpathy: Programming is “unrecognizable” — From 80% manual to 80% agent coding in three months

• Andrew Ng at Davos 2026 — “A person that uses AI will replace someone that doesn’t”

• 2026 State of AI Agents Report — 91% enterprise adoption, 57% running multi-step agent workflows

• Gartner: 40% of enterprise apps will embed agents by end of 2026

• Agentic Engineering Knowledge Map — 54-article synthesis on agentic patterns and architectures

• Context Engineering — The discipline of shaping what your agent sees, knows, and remembers

• Model Context Protocol — MCP: “USB-C for AI agents”

• Agents With Hands — Last month’s piece on agent capabilities

A lot more than I could have imagined, but also, as I pushed the boundaries, a lot less than I had hoped. Was it useful? Yes. Life-changing? Not really. The agent has access to your laptop (desktop) and a whole range of tools to “manipulate” your files. It may feel productive… but that’s the trap. Because the scary part isn’t whether it’s useful. The scary part is that tools like this quietly normalize a new behavior: granting an AI system real authority over your machine, often without “adult supervision.”

Don’t get me wrong, I use agents a lot. I’m just more comfortable if my agents are not freely roaming around. 😅

In this issue we’ll cover the popular AI agent OpenClaw, and the new releases of “assistants” like Cowork and Codex app—what are they, why do they matter, and most importantly, what now?

TL;DR (for skimmers who have a meeting in 4 minutes)

• Agents don’t just answer wrong, they can act wrong. Treat them like interns with root access.

• The easiest way to get burned is the “lethal trifecta“: private data + untrusted content + a way to send things out.

• The fix is not “don’t use agents.” The fix is cage them: least privilege, isolation, approvals, and a slow permission ramp.

Mark these words, and you can go enjoy your agents :D

P.S. https://agentcase.pioneeringminds.ai/ contains tutorials and examples for these tools : D

EastCoast AI Events

• Panel 02/09 Under the Hood of AI — Vanderbilt Hall, NYU

• Application 02/09 The New Landscape of AI-Driven Communication — Virtual

• Research 02/10 Teaching AI Ethics Through Project-Based Learning — Virtual

• Research 02/11 Scientific Research in the Age of AI — Virtual

• Application 02/11 How AI and Emerging Tech Are Shaping Business Transformation — Faculty House, Columbia

• Application 02/12 AI in Healthcare: Past, Present, and Next — Columbia Business School

• Research 02/13 What Can AI Teach Us About Student Reasoning? — Virtual

• Research 02/20 ML and AI Seminar Series: Tom Goldstein — School of Social Work, Columbia

• Application 02/21 AI Interfaces Hackathon with Claude — New York, NY

What Is an AI Agent, Really?

AI agents can take many forms with varying degrees of freedom. Frameworks like Google’s ADK, LangGraph, and many others use logical flows with “AI brains” embedded in the workflow. Think of an assistant who can really only help you schedule meetings—it can decide when and where the meeting might happen, but it cannot run your errands, clean up your house, or do grocery shopping for you.

Now, a different type of agent (which is really what the hype is all about today) is these terminal agents—pi-mono (used by OpenClaw), Claude Code, Gemini CLI, and more. I’ve always called these “general purpose agents“ since they can be used to do literally anything if they have the tools and context. Behind the scene, these are just brains on loop—we keep prompting LLMs to do tasks turn by turn. The implementation doesn’t get simpler than Mini SWE Agent if you want to dive into the particulars :D

You can see this pattern across the ecosystem:

• OpenClaw is a self-hosted, always-on assistant (pi-mono) that plugs into your messaging apps and runs tool executions: filesystem access, web browsing, CLI actions, and more.

• Anthropic’s Cowork packages Claude Code for non-technical users inside Claude Desktop. You describe an outcome, step away, and come back to finished work.

• OpenAI’s Codex app is the developer-flavored version: multiple agents in parallel, organized by project threads, with diffs you can review and built-in worktrees so parallel work doesn’t collide.

If you’ve been thinking “cool, a smarter chatbot”—not quite. This is closer to hiring an administrative assistant than installing an app.

Amazing Development… So Why Does This Change Everything?

To use these tools effectively, you give these agents access to all of your files and maybe even more privileges. For example, I gave mine access to WhatsApp (only allows messages from me though), access to my email, calendar and Google Drive, a local personal knowledge base, and basically admin access to my cloud services. This is extremely convenient—but can you imagine the risks involved?

The Failure Mode Is Different Now

Once an agent can touch files, run commands, browse the web, or message people, the risk shifts. It’s no longer “oops, hallucination.” It’s:

• “oops, it deleted the wrong folder”

• “oops, it pasted a secret key into the wrong place”

• “oops, it got tricked by a webpage and exfiltrated your data”

OpenAI’s safety guidance calls out prompt injections as a common attack vector: untrusted text enters the system and attempts to override instructions—sometimes triggering tool calls or leaking data. They’ve publicly discussed hardening their browser agent against this using automated red teaming.

Now layer on Simon Willison’s warning—the one I wish every agent UI forced you to read before the “Connect your Gmail” button:

If your agent combines private data access, untrusted content exposure, and external communication, you’ve built a data-exfil machine by accident. — Simon Willison

That’s the “lethal trifecta.”

This Is Not Theoretical Anymore—It’s Already Happening

Four signals from the last few weeks:

1. Moltbook (a social site for AI agents) exposed private messages, emails, and credentials for thousands of users according to Wiz, reported by Reuters—tied to “vibe coding” speed outrunning security basics.

2. OpenClaw CVE-2026-25253: A crafted URL could cause a client to connect and send authentication tokens—patched in version 2026.1.29.

3. Malicious skills: Researchers found add-ons uploaded to ClawHub designed to deliver malware or steal crypto credentials.

4. Cowork itself is marketed as a “research preview” with unique risks due to its agentic nature—Anthropic warns not to use it for regulated workloads.

If you’re thinking, “Okay, but that’s open-source chaos, enterprise tools will be safer…”—maybe. But the underlying pattern is universal: capability + autonomy + access creates a new attack surface.

The Lobster Cage Protocol 🦞

“Lobster Cage” is my favorite metaphor for OpenClaw-style agents: the claw is powerful… which is exactly why you keep it inside something sturdy.

This section is intentionally practical. You should be able to do this on a Sunday night and feel calmer on Monday morning.

Step 1: Pick Your Cage Level

OpenClaw’s own security guidance says: decide who can talk to it (identity), decide where it can act (scope), then assume the model can be manipulated. That ordering is correct.

Step 2: The One Checklist That Actually Matters

That’s the spine. You can do fancy stuff later.

Step 3: “Monday Morning” Starter Setups

To help you get started and explore what these agents can do, we have compiled community use-cases in

https://agentcase.pioneeringminds.ai/

as well as tutorials on how to set things up for each agent!

OpenClaw: Setting Up Your “Chat-Based Ops Intern”

Safety reminders for OpenClaw:

• Keep “web tools” off by default; enable only when needed

• Use a “reader agent” pattern: create a read-only agent for untrusted content (web, PDFs, emails), then pass summaries to your main agent

• Don’t install random skills from ClawHub like it’s an app store—the ecosystem has already seen malicious uploads

• OpenClaw has had serious security issues (now patched), so cage discipline isn’t optional “best practice”—it’s part of using the product responsibly

Cowork: The “Knowledge Worker in a Box” Setup

Best for: Cleaning folders, generating reports, wrangling docs, browser-based workflows—especially for non-devs.

Setup (5 minutes):

1. Create a folder called Cowork_Sandbox/

2. Copy in only what you want it to touch

3. Keep backups (seriously—Cowork can delete files)

4. When giving instructions, ask it to propose a plan first (”show me what you’ll do”), then execute

Anthropic’s own safety guidance explicitly recommends the dedicated-folder approach because Cowork can read, write, and delete files. Cowork is positioned as a research preview, and Anthropic flags the agent-specific risks (like prompt injection) more directly than most consumer tools.

Codex App: The “Parallel Engineering Line” Setup

Best for: Developers juggling multiple tasks, long-running changes, parallel exploration.

Setup principles:

1. Use worktrees by default—that’s the point. OpenAI’s docs describe worktrees as the way to run multiple independent tasks without collisions.

2. Keep permissioning intentional. Codex supports approval policies and read-only modes, and asks for approval when stepping outside the safe boundary (like network access).

3. Don’t start with 12 agents. Start with 2: one for tests + bugfix, one for refactor + docs. You’ll learn faster and review better.

The one-sentence explanation for why Codex app exists:

The challenge shifted from “can the model code?” to “can humans supervise multiple long-running agents without losing their minds?”

Final Thoughts

Agents with hands are real. They’re also not magic.

They’re powerful precisely because they blur the line between “advice” and “action.” That blur is where productivity comes from… and also where regret comes from.

So, when you actually use these types of agents (I personally recommend cowork for beginners), make sure you limit the directory within which agnets can work.

If you do just one thing after reading: create a dedicated folder (or a disposable VM) today. Then let the agent earn its way to more access.

References & Reading List

Anthropic. (2026). Getting started with Cowork. Claude Help Center. https://support.claude.com/en/articles/13345190-getting-started-with-cowork

Anthropic. (2026). Using Cowork safely. Claude Help Center. https://support.claude.com/en/articles/13364135-using-cowork-safely

OpenAI. (2026). Introducing the Codex app. OpenAI. https://openai.com/index/introducing-the-codex-app/

OpenAI. (2026). Worktrees (Codex app documentation). OpenAI Developers. https://developers.openai.com/codex/app/worktrees/

OpenAI. (2026). Security (Codex documentation). OpenAI Developers. https://developers.openai.com/codex/security/

OpenAI. (n.d.). Safety in building agents. OpenAI Platform Docs. https://platform.openai.com/docs/guides/agent-builder-safety

OpenClaw. (n.d.). Security (Gateway). OpenClaw Documentation. https://docs.openclaw.ai/gateway/security

Willison, S. (2025, June 16). The lethal trifecta for AI agents. Simon Willison’s Weblog. https://simonwillison.net/2025/Jun/16/the-lethal-trifecta/

Satter, R. (2026, February 2). ‘Moltbook’ social media site for AI agents had big security hole, cyber firm Wiz says. Reuters. https://www.reuters.com/legal/litigation/moltbook-social-media-site-ai-agents-had-big-security-hole-cyber-firm-wiz-says-2026-02-02/

The Verge. (2026). OpenClaw’s AI ‘skill’ extensions are a security nightmare. https://www.theverge.com/news/874011/openclaw-ai-skill-clawhub-extensions-security-nightmare

SecurityWeek. (2026). Vulnerability Allows Hackers to Hijack OpenClaw AI Assistant. https://www.securityweek.com/vulnerability-allows-hackers-to-hijack-openclaw-ai-assistant/

Happy New Year, Pioneering Minds!

If 2025 was the year you tried AI, 2026 is the year you build small AI workflows that actually stick, in your work, your learning, and your life!

In 2026, we’ll help you become an AI Pioneer by building small, repeatable AI workflows. In Today’s issue: 3 reflections that shaped my thinking + a starter project you can begin this week.

In 2025, we hosted 20 events with partners across startups, law firms, and universities and non profits - a special thanks to Cooley who have been providing us with legal advisory! A sincere thanks to every partner, speaker, and attendee of 2025! You have been an amazing part of the community!

**Full sponsor/partner list at end of article.

Running into 2026, our focus is simple: more practical value. This means more hands-on events, more guided learning, more practical AI examples and more ways to learn with mentors - detailed below!

What happened in 2025?

Now, did you get the year-in-review from ChatGPT? I sure got mine! It reminded me of something true: AI isn’t a tool I “use sometimes” anymore. It’s part of how I learn, plan, and build. It has become part of how I live.

Over the past two weeks, I read three end‑of‑year reflections on building with AI—one for builders, one for founders, and one that’s more philosophical. All three landed on the same message: AI amplifies execution, but humans still own direction.

• For builders: Boris Cherny on how he uses Claude Code

• For founders: Tao Zhang (Manus) on what he believed before building Manus

• For thinkers: Dr. Yuandong Tian (田渊栋) on finding value “above the flood line” in an age of abundant AI

original content in chinese so I created slides with notebookLM for ya

Boris’ post is the most immediately actionable: treat AI coding like working with a junior teammate: break tasks down, give tight specs, and review the output relentlessly.

Manus’ core idea is simple: agents don’t become useful when they’re clever—they become useful when they’re persistent. If your AI “forgets everything” each session, it’s just not even a toy, it’s unsable. The product path that wins is: start broad, learn from real usage, then turn the few workflows people repeat into reliable defaults.

Dr. Tian’s point hit me hard: in a world of “abundant genies,” the advantage isn’t getting answers, it’s choosing a destination and sticking with it. As AI makes execution cheaper, strategy becomes everyone’s job, not just leaders’.

In other words: if AI becomes your teammate, your job shifts to setting direction, delegating clearly, and reviewing quality. That’s the core skill of an AI Pioneer. These were only required for the leaders before, but now, now you have ateam and YOU are becoming the leaders of AI!

Next up, build something for yourself!

If this got you fired up to build something small for yourself, good. Because here’s the point of 2026: we’re going to practice. I’ll share one starter project below, but first, here’s the learning loop I use to ramp up fast:

• ChatGPT 5.2 Thinking, Gemini to do deep research

• (occasionally) manus to do wide research for comprehensive research

• NotebookLM to help generate easy to digest content

• Notion to store the content and summarize with notion AI

• A pen and notebook to write down my thoughts and of course my phone to take photo and ask GPT/Gemini to answer my questions on the note!

If you try one thing this week: use AI to learn one topic you care about for 30 minutes (or 2 hours, or however long you like!). Then, write a one page summary in your own words. This combo helps solidify the knowledge you’ve just gained.

Starter Project: Your Personal Information Summarizer (that learns your taste)

One project you can start small and make more powerful over time: a personal information summarizer (not “news”) that gradually learns what you actually care about.

Why this is a great first personal AI tool: it teaches you the difference between retrieving information and building taste. Retrieval finds what you saved. Taste learns what you consistently value and filters the rest.

What it does (simple version)

• Pick a few sources (newsletters, papers, YouTube channels, podcasts, saved links)

• Summarize what’s new

• You give quick feedback (“more like this / less like this” or just plain sentence)

• Over time, it becomes a taste filter, not just a summarizer

You can eventually add RSS feeds, memory, and automation. But don’t start there.
Start with the simplest version: collect sources → summarize → save/ignore → reflect weekly.

Then, ask ChatGPT or Gemini how to expand it: rss feed (which is used to get updates about news, journal articles, videos, pod casts…all kinds of message sources), memory management (agentic rag), ai content summarizer. Personalize summarization and recommendations… all the fun stuff!

The best and most valuable aspect of this project is that it creates something uniquely you and helps you curate your content your way. And gets you familiarized with the AI development techniques and practices.

Pioneering Minds AI In 2026

Even after using AI daily for two years, I still learn something new constantly. The pace is real, so in 2026 we’re building structure around it.

1. Community Platform and Mentors

To help you become the AI Pioneer who can not just survive in the AI era, but also thrive in the new era, we are launching our community platform in early Q1.

A place where you can find your peers to exchange ideas and insights around AI, share your AI projects, learn from high quality courses and practice AI with real challenges like the one I shared above with instructions and help from experienced AI mentors within the community!

Mentors are currently accessible at the main website and we are in the process of onboarding more mentors, stay tuned!

P.S. Join the waitlist at pioneeringminds.ai to receive the newsletter and a platform invite when we launch!

2. More Events for Builders and Founders !!

We are hosting more curated AI events to support our professionals and founders through 2026! A public roadmap will be published again at www.pioneeringminds.ai in the coming week!

If there’s an event you want to see, or a topic you’d like to lead, reply and tell me what you’re building or learning.

3. New 2026 Annual Sponsor Program

Our Promise to Members

• No spammy sponsorships.

• No random ads.

• Only partners that are genuinely useful to our audience.

As a nonprofit, our team runs on volunteer effort. If you’ve found value in this community, sponsorships and donations help us keep events and content high-quality.

Last year we learned something important: sponsorship shouldn’t be “one event visibility.” The best partnerships show up consistently!

So for 2026, we’re offering a full-year sponsorship package: sustained visibility across our newsletter, events, and community rather than a one-off logo slide.

Why annual beats single-event sponsorship:

• Consistent presence all year (newsletter + events + community)

• Better alignment with a focused audience

• Higher trust and ROI through repetition

If you want the sponsor packet or want to explore fit, reply to this email or contact me at nickgu@pioneeringminds.ai.

Wishing you a focused, creative, and powerful 2026.

Dr. Nick Gu

Founder & President, Pioneering Minds AI

📅 January 2026 Events

• Agentic AI in the Trenches | Startup Grind | Jan 6, 6–9 PM | In-Person | $5 | RSVP

• Agentic AI Summit | ODSC | Jan 21–Feb 5 | Virtual | Free–$299 | RSVP

• Tech Talk w/ Glean, FlutterFlow & Alaffia | NY AI Engineers | Jan 21, 6–8 PM | In-Person (Brooklyn) | Free | RSVP

• Responsible AI: What Leaders Need to Know | Brown | Jan 22, 12–1 PM | Virtual | Free | RSVP

• Overcoming AI Deployment Challenges | MIT xPRO & CSAIL | Jan 27, 11–11:45 AM | Virtual | Free | RSVP

• AI in Education: Open Questions Panel | AiiCE | Jan 27, 5–6 PM | Virtual | Free | RSVP

• ORBIT Winter School: Real-World Data Analytics | Duke DCRI | Jan 28–30 | Virtual | Free | RSVP

• Career Conversations: AI/ML Careers | NYU Wasserman | Jan 28, 5–6:30 PM | Virtual | Free | RSVP

• A Geometric Perspective on LLM Behavior | Princeton CITP | Jan 27, 12:15–1:15 PM | Virtual | Free | RSVP

• AI & Tech Mixer NYC | NYC AI & Tech Meetup | Jan 29, 6–9 PM | In-Person (Brooklyn) | Free–$54 | RSVP

• AI+Aging Seminar: SpeechCARE | Columbia Aging Center | Jan 29, 1–2 PM | Virtual | Free | RSVP

📚 References & Reading List

1. Clear, J. (2018). Atomic Habits: An Easy & Proven Way to Build Good Habits & Break Bad Ones. Avery. https://jamesclear.com/atomic-habits

2. Ribeiro, M. T., Singh, S., & Guestrin, C. (2020 ). Beyond Accuracy: Behavioral Testing of NLP Models with CheckList. Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics. https://doi.org/10.18653/v1/2020.acl-main.442

3. Liang, P., et al. (2022 ). Holistic Evaluation of Language Models. arXiv preprint arXiv:2211.09110. https://arxiv.org/abs/2211.09110

4. Datadog (2025) “Building an LLM evaluation framework: best practices”

Sponsors and Partners of 2025

• Law Firms

  • Cooley

  • Polsinelli

• Non Profits

  • 1 Thing Org

• Startups

  • TRAE, Lastmile AI, Relay.app, Hey Hunchi, DeepL, ArcSphere, aigensei, ADP, Xenos Wealth, behamics, Nachonacho, Superlinear Academy, Sanctum AI

• Academic institutions and groups

  • NYU CEC, Northeastern University (and EAI), Iona Univerisity.

Join us for AI Hacks Back: Founder Risks, a live panel on how AI is transforming offensive cybersecurity, and what that means for founders today. Cybersecurity and legal experts will break down real-world attacks like large-scale prompt injection, agentic intrusions, “vibe coding” vulnerabilities, and AI-accelerated zero-days, with a focus on what founders can do now to defend at machine speed. Designed for builders, operators, and technical founders, this session will give you a clear, practical view of the risks already emerging in 2025 and how to prepare for 2026. Stay tuned to our newsletter and Luma calendar for event details and registration.

🌟 Featured Events

🤖 Application

• Dec 2: Cornell Tech: Builders and Disruptors - Reggie Fils-Aimé (ex-Nintendo) on innovation in AI era

• Dec 9: Intro to AI and Vibe Coding Tools Workshop - Hands-on with Lovable & v0. with experts from NYU, Meta & Google

• Dec 9: The Future of AI Infrastructure (Oracle & NVIDIA) - Deep dive on GPU orchestration at cloud scale

• Dec 10: UGLY TALK: AI for Ecommerce Logistics - Operators from large-scale ecommerce + Acronym VC sharing actionable AI workflows

🛠️ Deep Dives

• Dec 2: XPENG VLA2.0: Foundation Model for Physical AI - Industry’s first mass-produced billion-parameter physical AI model

• Dec 2: Mel Andrews: False Promises of Fair ML - Princeton CITP talk on algorithmic fairness limitations

• Dec 3: MIT: How Curiosity Shapes Well-Being - Technology and learning discussion

• Dec 3: Asia’s Strategic Edge in AI - Nikkei @ Asia Society

• Dec 4: Embodied Ethics: What Dance Teaches Us - Northeastern RAI workshop on moral imagination in tech design. No dance experience needed!

🎭 Networking & Social

• Dec 2: NachoTuesday: SaaS & AI Holiday Mixer - Founders, VCs, tech media. Application required, carefully screened

• Dec 3: NYC AI Users Happy Hour - AI enthusiasts mixer featuring favorite former speakers

🌟PMAI X TRAE Hackathon!

Pioneering Minds AI partnered with TRAE to host the TRAE SOLO Hackathon NYC, and it was a blast of a day. Over one packed day, founders, engineers, and builders came together to create projects centered on Reliable and Responsible AI.

Powered by TRAE SOLO, the next-gen AI IDE, teams moved from concept → code → deployment with impressive speed. With full plug-in support and API connectivity, participants could integrate external systems, call services directly, and extend the environment with custom agents and tool-chains, turning ambitious ideas into working prototypes in hours, not weeks.

Projects spanned tourism, AI evaluation, fintech, health tech, and more, showcasing both technical depth and a strong sense of purpose.

Thank you to Trae for partnering with us, and to all the mentors, judges, and participants who brought such heart and momentum to the room.

Events like this reflect why we started Pioneering Minds AI: to empower builders to create AI that’s not just powerful—but responsible.

𖠌 Grab Time With Our Mentor & Advisor!

For those of you desire to get more hands on support and mentorship to learn about AI, please grab time with our President, Dr.Nick Gu at https://www.pioneeringminds.ai/mentors !

We are actively onboarding more mentors in the coming months, so stay tuned! If you would like to give back to the younger generation by offering your expertise and insights, please email team@pioneeringminds.ai

🔮 Final Thoughts

Even though December is the holiday season, NYC is still packed with solid AI events for those staying around!

2025 was the year vibe coding went mainstream and AI tools became genuinely useful for builders. The gap between “idea” and “working prototype” has collapsed from weeks to hours. But here’s what I keep thinking about: the tools don’t matter if you’re building the wrong thing. Speed without direction is just faster failure.

As you head into the holidays, I’d encourage you to try one thing: Build something small using the vibe coding tools. A tool for yourself. 15 minutes. Just to feel what’s possible now - and share in comments below!

See you in 2026, pioneers. We’ve got a lot more to build together.

Happy holidays! 🎄

Building the future, one product at a time.

The Pioneering Minds Team

Hello our AI pioneers!

Do you want to discuss AI but couldn’t find your group? Do you want to build more but lacks ideas? Do share products and projects you’ve built with your peers? Do you have challenges you want to solve but don’t know how? Are you looking for jobs? Mentors? Investors?

What if I tell you, our Community Platform can help you to find the peers, mentors and resources you are looking for and grow yourself in the process?

Thanks to our exceptional team We built the platform to further our organization’s mission

Grow AI-Native Pioneers—professionals, founders, and scholars—who build, deploy, and govern AI that compounds human opportunity responsibly.

And it’s now open for beta access! Join the waitlist today! We want to make this as light hearted as possible, so as more pioneers sign up, we will unlock increasingly valuable rewards for raffle and resources for all members!

Cheers,

Nick Gu

President, Pioneering Minds AI

🌟 Featured Events

• TRAE SOLO Hackathon: Build Reliable AI | Nov. 22, 9 AM-6:30 PM, Full day Hackathon session for audience of all backgrounds, with hands on workshop RSVP

• Agents in Production - MLOps x Prosus | Nov 18, 2-8:30 PM, Virtual event with 30+ talks on deploying AI agents in production RSVP

• The AI Advantage: Building and Funding Startups | Nov 14, 10 AM-1:30 PM, CBS Startups Week conference with founders and investors RSVP

📅 November Events: Your Complete NYC AI Calendar

November brings an unprecedented concentration of AI events to NYC. We’ve organized them by focus area to help you strategically plan your month.

🔧 Pioneering Minds Events

• TRAE SOLO Hackathon: Build Reliable AI | Nov. 22, 9 AM-6:30 PM, Full day Hackathon session for audience of all backgrounds, with hands on workshop RSVP

🤖 Application

• November Demo Day ft. Google Cloud | Nov 17, 6-9 PM, Showcase your latest AI demos and dive deep into LLMs with NYC’s best technical community RSVP

• Unleash Your Productivity with Gemini AI | Nov 24, 3-4 PM, Comprehensive guide to leveraging Google’s Gemini AI for streamlining workflows RSVP

• Responsible AI for Human Well-Being | Nov 11, 4-5:30 PM, Focus on wearables and digital healthcare AI applications at Stevens Institute RSVP

• Agents in Production - MLOps x Prosus | Nov 18, 2-8:30 PM, Virtual event with 30+ talks on deploying AI agents in production RSVP

• The AI Advantage: Building and Funding Startups | Nov 14, 10 AM-1:30 PM, CBS Startups Week conference with founders and investors RSVP

🛠️ Deep Dives

• 2025 UCL NeuroAI Annual Conference | Nov 5, 9:30 AM-5 PM, UCL conference bringing together neuroscience and AI researchers with keynote speakers RSVP

• Bias in AI Reading Group | Nov 5, 11 AM-12 PM, Discussion on fairness issues in AI systems (Online) RSVP

• AI/ML Conversations Meetup: Smart CPU Offloading | Nov 12, 5:30-7:30 PM, Monthly meetup on scalable LLM inference at Capital One Office RSVP

• Panel on Building AI Responsibly | Nov 17, 6:30-9 PM, Discussion with Code with AI founders on safety-first AI development RSVP

🎭 Networking & Social

• NYC AI Users - AI Enthusiasts Happy Hour | Nov 6, 7-10 PM, Social mixer with favorite former speakers at Smithfield Hall RSVP

• Post-Reality: A Prototype from the Edge | Nov 10, 7-9 PM, Interactive storytelling experiment at Film at Lincoln Center RSVP

• The VC Playbook: Nailing Your Pitch, GTM, and AI Ops | Nov 11, 6-9 PM, Fireside chat on building scalable AI businesses with ff Venture Capital RSVP

💡Use AI and Keep Your Data Safe

In last episode we discussed the mentality to approach working with AI. Since then, there has been quite a few launches from big players - Google’s Upgraded AIStudio, OpenAI’s Atlas, Claude Code Skills for example. However, with all of these products, we are rushing into an era of increased risk of data exploitation. So… How we can get productivity gains while keeping our information safe?

Busting Myths: LLMs and Your Data

Many misconceptions are floating around about how large language model (LLM) tools handle your data. Let’s debunk a few key myths:

Myth 1: “Consumer models always learn from everything I type.”

Reality: Not anymore. All major consumer AI providers—OpenAI, Anthropic, and Google—now offer explicit controls to disable model training on your conversations. For ChatGPT, you can turn off “Improve the model for everyone” in your Data Controls. For Claude, you can disable training in your Privacy Settings. For Gemini, turning off “Keep Activity” prevents your chats from being used for training. However, be aware that defaults vary. As of October 2025, Anthropic’s Claude is opt-in for training, while others may be opt-out.

Myth 2: “Privacy features make my data completely safe.”

Reality: “No training” does not mean zero retention. Providers still retain data for a period to monitor for abuse, fraud, or legal reasons. For example, OpenAI’s “Temporary Chats” are deleted after 30 days. Google Gemini saves chats for 72 hours even with activity history off. Furthermore, using plugins, browsing, or other connected tools can introduce third-party data paths with their own policies.

Myth 3: “Enterprise and API accounts have the same privacy risks as free accounts.”

Reality: This is false. Enterprise and API services from OpenAI, Anthropic, and Google Cloud operate under a fundamentally different, more protective privacy model. By default, they do not train on your business data. These services are governed by commercial terms and often come with a Data Processing Addendum (DPA), providing stronger legal and technical protections. If you are handling sensitive business data, upgrading to an enterprise or API plan is a critical step.

Your Task Today: Disable Data Sharing For Your apps

Take a look at all the AI products you use, and disable data sharing explicitly in the App settings.

You can build an AI Agent to help you figure out how to disable data collection for AI products using n8n. Below is a small example

• Privacy Agent Chat

• Workflow on Github

This agent was built on n8n, using Gemini API and Serp API for google search. This is the simplest format of an AI Agent - Chat, Model, Memory, Tool.

We have disabled the execution, meaning that your execution will NOT be saved by us. So feel free to chat & explore how to ‘safe guard your data’ here! The prompt was generated with GPT-5 Thinking Extended + Websearch. You can easily copy this and create your own n8n workflow : D

[Advanced] Safe (and Reliable) Research with AI Agents: Two Smart Approaches

Finally, let’s look at how you can harness AI as a research assistant safely. We’ll explore two scenarios: (A) using a conversational AI like ChatGPT itself to help gather information, and (B) creating your own custom “agent” with API calls. In both cases, the goal is to get the answers you need without leaking your data.

A. Researching with Public AI Product

Using public AI services (e.g. ChatGPT, Claude, Google’s Gemini) for research is safer than it might sound. Major providers state they will not train on your inputs (AFTER you opt-out). These companies may retain your chats for a short period solely for compliance/safety (OpenAI API logs auto-delete after 30 days), but essentially your content won’t be mined into their models without consent. In short, large AI platforms won’t misuse your data – and many offer enterprise plans with even stricter privacy guarantees.

So how do you actually do deep research with ChatGPT or Gemini? Here’s a brief guide:

1. Define Your Question and Plan: Start with a clear research question or hypothesis. Break it down if it’s complex - and don’t hesitate to use AI for this step.

2. Use Advanced Models & Features: Leverage models with reasoning capacity and searching grounding for literally all topics of value. Advanced tiers have bigger “brains” for reasoning and are less likely to hallucinate.

3. Iterative Querying – Refine as You Go: Deep Research products usually requires iterative planning and confirmation. You would want to add your clarifications and comments during this step to shape the model behavior.

4. Verify and Cross-Check: Always validate critical facts. AI can occasionally get things wrong or outdated. If your AI provides sources or citations, check those links to confirm the info. The goal is to use AI’s speed to gather info, but use your judgment to verify it.

Above is an example of iterating on the prompt to create a comprehensive deep research. The generated result (exported to Google doc) shows a decent amount of references for the requested topic.

B. Building a Custom Deep Research Agent

For even more control, consider building your own research agent using AI APIs. This approach lets you customize how the AI works (and even run it entirely locally), which can be both powerful and privacy-preserving.

Due to the length of the content, we can’t cover all details in this issue. We will post example of this in the community platform soon.

Single vs Multi-Agent: First, decide if you need one AI agent or many. A single-agent system means one AI model handles the entire task sequentially – it reads the question, searches for info, composes an answer, etc. This is simpler to build and works well for many cases. However, Anthropic’s research found that a multi-agent setup can supercharge complex research tasks. In their internal tests, a team-of-AIs approach (one “Lead” researcher AI directing several sub-agents in parallel) delivered 90% more output than a single Claude model alone. This divide-n-conquer technique has been used in product like Manus as well and can be re-created on your local machine.

Using an existing framework: Instead of creating DR agents on your own, you can easily leverage existing examples, such as Deep Research Agent provided by Google, Open Deep Research by LangChain, DeepResearchAgent, or Deerflow. Each of these can be easily set up locally (tip: use claude code to help you with this)!

Switching to a Local Model: What if you have ultra-sensitive data and don’t want to call any cloud API at all? The good news is you can run these agents with a local Large Language Model. Projects like Ollama make it easy to host LLMs on your own machine and expose them via an API. You can simply hook the previously developed agent, and connect those to your local API. This blog has a detailed walk through.

In summary, AI can be a powerful partner without compromising your privacy – either by trusting reputable providers who protect your data (and opt-out), or by taking the reins and hosting the intelligence yourself.

Building the future, one product at a time.

The Pioneering Minds Team

🎯 Key Insights

1. AI Education Becomes Essential

As AI capabilities explode, the gap between those who can leverage these tools and those who can’t widens daily. Our new whitebook addresses this head-on with structured pathways for every level—from beginners to enterprise architects.

2. Memory is the Key to Autonomous AI

Long-term memory transforms AI from stateless tools to persistent collaborators. With breakthroughs in AI-native memory, RAG systems, and temporal knowledge graphs, agents now retain context across sessions, learn from interactions, and build genuine expertise over time. OpenAI’s ChatGPT memory for Pro users and enterprise deployments of LangGraph + MongoDB are making conversational continuity the new baseline expectation.

3. Responsible AI Cannot Be an Afterthought

With EU AI Act enforcement beginning August 2025 (fines up to €40M or 7% of revenue), responsible deployment is now mandatory. Red teaming has evolved from security theater to systematic evaluation—CISA’s TRAINS Taskforce, Japan’s AISI framework, and industry leaders like Anthropic are establishing continuous testing protocols. AI guardrails now span hallucination prevention, bias detection, and compliance validation, with 57% of consumers ready to switch brands over AI trust concerns.

🌟 Featured Events

• Pioneering Minds AI Demo: Founders Ask | Oct 10, 6-7:30 PM Interactive session where founders get real-time feedback on challenges (Free for founders/investors) RSVP [Speaker Apply]

• AI for Educators Summit | Oct 20 | Hands-on, accredited summit for K-20 educators with practical AI integration training and AI Readiness Certificate. RSVP

• Super AI ML Summit NYC | Oct 30 | Partnership with NYC Mayor’s Office bringing together the brightest minds in AI/ML for insights from operators, investors, founders, and policymakers. RSVP

• (IRL) Golden Garden Hour Gala | Oct 10, 6-9 PM FAANG++ and Ivy++ networking with themed sections and activities RSVP (CODE: PMAI)

📚 ANNOUNCEMENT: AI Education Whitebook Launch

We’re thrilled to release our comprehensive AI Education Whitebook—your roadmap to navigating the AI revolution!

What’s Inside:

3 Learning Paths:

• Beginners: Math → Python → ML → Deep Learning → Deploy in 30 days

• Developers: LLM development, AI agents, prompt engineering, production deployment

• Leaders: Implementation roadmaps, ROI frameworks, team upskilling strategies

Exclusive Resources:

• 50+ courses from MIT, Stanford, Harvard (many free!)

• Personalized learning track generator

• Monthly tool updates & expert insights

📖 Get the FREE Whitebook!

📅 October Events: Your Complete NYC AI Calendar

October brings an unprecedented concentration of AI events to NYC. We’ve organized them by focus area to help you strategically plan your month.

🔧 Pioneering Minds Events

• AI Demo: Founders Ask | Oct 10, 6-7:30 PM Interactive session where founders get real-time feedback on challenges (Free for founders/investors) RSVP

🤖 Application & Development

• NYC October Demo Day | Oct 2, 6-9 PM Show your latest LLM projects and see cutting-edge demos from Google, OpenAI engineers RSVP

• Augmenting Talent from Within | Oct 15, 1-1:45 PM (Virtual) Tom Davenport keynote on organizational AI transformation RSVP

• Human-Centered AI | Oct 21, 1-1:45 PM (Virtual) Building AI products users actually trust and adopt RSVP

• Smarter Projects & Stronger Leaders | Oct 27, 6-8 PM Columbia’s M.S. in Project Management program launch RSVP

🛠️ Deep Technical Dives

• Agents at Work | Oct 9, 6-8 PM Columbia Business School talk by Prof. Namkoong on production agentic systems RSVP

• Disability & Access in Tech/AI Summit | Oct 9-10 Hybrid summit on building inclusive AI systems RSVP

• Narrative 2 Numbers Hackathon | Oct 11, 8:30 AM-6 PM Columbia’s interdisciplinary hackathon for public health data innovation RSVP

• AI Networking Summit 2025 | Oct 22-23 Two-day conference on enterprise AI infrastructure ($599-1,699) RSVP

• Urban AI Symposium | Oct 28, 10 AM-5 PM NYU Tandon’s exploration of AI in urban planning and smart cities RSVP

• Super AI ML Summit NYC | Oct 30, 9 AM-5 PM Major summit with NYC Mayor’s Office, featuring Fortune 500 and startup leaders RSVP

🏥 Specialized Domains

• AI in Public Health | Oct 8, 6-8 PM NYU panel on AI transforming healthcare delivery and medical education RSVP

• AI for Educators Summit | Oct 20, 9 AM-5 PM Full-day training for K-20 educators with certification opportunity RSVP

• AI+Aging Seminar | Oct 30, 12-1 PM (Virtual) Dr. Peter Abadir on AI innovations for healthy longevity RSVP

🎭 Networking & Social

• Golden Garden Hour Gala | Oct 10, 6-9 PM FAANG++ and Ivy++ networking with themed sections and activities RSVP

• AI Founders Supper Club | Oct 15, 6-9 PM Curated dinner for 15-20 early-stage AI founders (referral-based) RSVP

🚀 Key AI Releases (August-September 2025)

🧠 Smarter AI Models

What’s New: OpenAI’s GPT-5 (Aug 7) and Alibaba’s Qwen3-Max (Sept 24) represent the next generation of AI that can adapt their “thinking time” based on task complexity.

Why It Matters: These models work faster on simple tasks and slower on complex problems—like having an AI assistant that knows when to give quick answers versus when to think deeply. GPT-5 is already available in ChatGPT for all users.

🤖 AI in Your Workplace

What’s New: Microsoft Copilot 365 added multiple AI options (Sept 24), letting users choose between different AI models. DeepSeek V3.1 (Aug 21) introduced user-controlled reasoning depth.

Why It Matters: Your workplace AI tools now offer more flexibility—switch between AI models like choosing between different consultants based on their strengths. This means better performance for specific tasks like research or content creation.

🎨 Creative AI Tools

What’s New: Three major creative tools launched: Google’s Gemini image editor “Nano Banana” (Aug), Luma’s Ray3 video generator (Sept 18), and Meta’s Vibes AI video platform (Sept 25).

Why It Matters: Create professional videos and images without technical skills. Ray3 integrates with Adobe Firefly, while Meta Vibes lets you share AI videos directly to Instagram and Facebook—perfect for marketing teams needing quick content.

💡Not a tool, but a co‑worker – Mentality For Thriving in the AI Era

Create frameworks and systems to leverage AI (Agents) efficiently, just like how you would manage a team.

Help! This tool doesn’t work…

Many of you may have already leveraged AI agents to various degrees. Did you ever notice the AI ducking work, giving you false information or promising things it never delivered? If so, then congrats – you’ve had your first taste of managing your employees… or AI employees to be precise. These issues with AI are not new. Many of you have already observed them while using simple chats. The more autonomous the AI products are, the greater the problem, because every mistake in the decision chain amplifies the next one.

We run into trouble when we treat AI as if it were a calculator. Unlike deterministic tools where everyone gets the same experience once installed, large language models’s behavior could vary significantly based on how it’s used. Every piece of context you feed into a model biases its responses [1], so what you get from ChatGPT can be vastly different from what someone else gets.

Sounds familiar? If you’ve hired interns, have you ever met two interns with the exact same personality who can do the exact same things with the exact same quality? The tasks we assign to AI – programming, research, answering complex questions – are inherently ambiguous and delivered through natural language. The uncertainty in its output isn’t a failure of the model; it reflects the complexity of the problems it’s asked to solve. As soon as you see AI this way, the frustration starts to make sense.

At this point you probably understand where I’m going – in the era of AI, we shouldn’t treat the AI tools simply as tools, but rather as our co‑workers or employees who need help, guidance and support to grow into “someone” competent enough to perform the task. An intern analogy works nicely here: you wouldn’t expect every piece of data from your intern to be perfect, you’d build trust over time and learn their strengths and weaknesses. That’s exactly how you work with AI – start by double‑checking its work, then delegate more freely as it proves itself. Your expectations and the feedback you give shape its performance; psychologists call this the Pygmalion effect [2]. To simply put - the higher the bar you hold for your tools, the better it becomes.

Why does it matter?

There are several layers of AI products today. Simple turn‑based AI chats perform limited actions, multi‑turn AI agents can execute more complex tasks such as deep research or agent mode, and finally we have swarms of agents working together to achieve complex goals at scale. As the autonomy of AI tools grows, so does the necessity to manage them properly. Working with chat models, you need to provide context, roles, output requirements and a clear task – this was called prompt engineering. At its core, we’re biasing the model toward the behavior we want. These are just “brains” that can sense but not act.

Working with an AI agent, you again need to provide context, roles, output requirements and a clear task. However, unlike before, these agents can act – as long as you provide them with the tools. And guess what? You need to provide even more “training” so it can leverage those tools effectively. Managing a single AI agent is like managing a human. I recall when I first started my career: my manager gave me a doc to read and told me to find those familiar with the topic, hash out the details and implement the solution - which is now serving millions of users today but it all started with a simple doc. He wasn’t breathing down my neck – he gave me autonomy. He also made sure I had the necessary context and support: points of contact I could talk to, docs I could read and lessons from previous decisions. He was engineering the context I needed to get the job done. Today we’d call this context engineering - for AI.

When you design tasks for your AI “employee,” think like a human resources pro rather than an engineer. Classic job design research reminds us that people value autonomy, variety, understanding the significance of their effort and getting feedback on specific tasks. The same principles hold when you frame tasks for an AI. Instead of micromanaging every keystroke, break your project into meaningful chunks and explain why each step matters. Vary the work and provide frequent feedback loops. You might even let the model choose its own tools within limits. In my experience, AIs that have room to decide and see the impact of their work produce better results and require less hand‑holding.

To put simply, my single-agent AI works much better when I’m serving as a Tech Lead - providing guidance and directions, instead of micromanaging every single aspect of it’s work. This is only going to be more true as the foundational model evolves.

Scaling up: from a single agent to swarms

Working with swarms of agents – not just one or two or three, but five, ten, twenty or more – is much more than a single person can manage. I put my Claude code to work 24/7 at one point, which translated to six to ten parallel sessions running, or six to ten agents working. Now I run much more… and, just like any organization, the pain of scaling became obvious as the work scaled up. I was unable to keep up with the work generated, unable to keep track of where things were happening. All of these problems surfaced as I scaled my operations.

If this sounds like a human company outgrowing its first manager, that’s because it is. As organizations grow, leaders introduce hierarchy: the manager becomes a senior manager and hires first‑line managers to lead smaller teams. The senior manager stops reviewing every line of code or every document comes along, and instead becomes an architect of decision-making systems - designing guardrails rather than steering wheels, scorecards rather than spot-checks. The same approach works with AI. When your “team” of agents becomes larger than what you can personally supervise, introduce layers: designate certain agents to monitor and verify outputs or build automated checks to catch obvious hallucinations. Your role shifts from doing the work to designing workflows and processes.

This isn’t a problem; it’s an opportunity. A handful of AIs can have the output of three to five engineers. When you manage several AIs in parallel, you’re effectively leading a team of a dozen or more. In traditional careers, becoming a middle manager overseeing ten people requires technical skill, management prowess and luck [3]. With AI, you can have a team of equivalent size, ready to work 24/7 with exceptional memory transfer. If your management skills can keep up, this is a resource that was incredibly scarce in the past. Rather than micromanage, focus on high‑leverage activities: set clear goals, design the system of work and invest in tooling that reinforces good practices.

What’s next?

Using chat – manage context. Using an agent – be patient and give it support. Using swarms of agents – make sure you have a clear goal and purpose for the agents. Because AI problems are really human problems in disguise, the best practices we’ve developed for hiring, motivating and evaluating people apply here too. Craft an honest “job description” for your model, have regular check‑ins instead of annual appraisals, and provide feedback in a psychologically safe way. Above all, keep your mental frame rooted in management: you’re not pressing pedals any more; you’re navigating. This is a new era, and I would encourage readers – many of you are senior leaders in your respective fields – to think about what you’ve done over the years and how your usage of AI can be influenced. We welcome readers to submit questions and thoughts!

[1] Dong, Q., Li, L., Dai, D., Zheng, C., Ma, J., Li, R., ... & Sui, Z. (2022). A survey on in-context learning. arXiv preprint arXiv:2301.00234.

[2] Feldman, R. S., & Prohaska, T. (1979). The student as Pygmalion: Effect of student expectation on the teacher. Journal of Educational Psychology, 71(4), 485.

[3] grapeot, (2025), Managing AI: The Most Important Promotion of Your Career, https://yage.ai/ai-management-2-en.html

🔮 Final Thoughts

The Compound Effect of AI Mastery

Every week you wait to build your AI capabilities is a week your competition pulls ahead. The tools released in September alone—from reasoning models to autonomous agents—represent years of traditional software development compressed into single API calls.

But tools without knowledge are just expensive toys. That’s why we created the AI Education Whitebook: to give you the structured path from wherever you are to wherever you want to be in AI. Combined with October’s incredible lineup of events and the intelligence system tutorial above, you have everything needed to level up.

The future belongs to those who can orchestrate AI systems, not just use them. Whether you’re building products, conducting research, or making strategic decisions, the ability to create custom AI workflows will become as fundamental as using spreadsheets is today.

Three Actions to Take This Week:

1. Download the Whitebook and identify your learning path

2. Register for 2-3 October events that match your goals

3. Start building your intelligence system with our template

Remember: In the AI era, the gap between learning and doing is disappearing. Every tutorial is executable, every model is accessible, and every idea can become reality in days, not months.

See you at the October events! 🚀

Building the future, one product at a time.

The Pioneering Minds Team

💡 Editor's Note: AI itself has evolved so much more these past few months. The launches from claude code, to gemini cli, to chatgpt codex/agent mode, cursor mobile… All the tools you can think of are their - the pieces to create a optimal life is already here. Now we just need to put them together - with the help of AI no less. If you ever wonder how you can improve your workflows, learn faster and be more productive, then check out our resources and tips (definitely not reading assignment 😉) - hopefully it helps! Let’s dive in!

🚀 New Program Launch

To further our mission, we are proud to unveil the AI Mentor & Advisory Program!

As global accelerators—from Google’s “AI First” initiative for startups to GMI’s AI‑enhanced mentorship model—show the catalytic power of expert guidance, we’re bringing that momentum to purpose‑driven founders.

If you’re building a startup, integrating AI into your business, boosting your team’s productivity and want hands‑on AI expertise, apply now to join our inaugural cohort!

🎯 Our Upcoming Events (Quick Card)

• Building the First Mobile-Native AI Browser
  Thu, Aug 7 | 17:30-19:00 EST | Register →

• AI For Reals – Deploying AI in Your Business
  Thu, Aug.21 | 17:30-19:00 EST | Register →

If you are few of the folks who are staying at home during August - we’ve got AI events scheduled to keep you busy!

📚 Learning Resources

We have distilled a list of AI resources that helps you understand what is AI and how to use it. The world doesn’t need another long list - we are all overload with information these days. So we keep it short!

For those of you who enjoy long white books, we have one in the cooks as well & will release in coming month!

For Technical Learners: From Foundations to Frontier

1. Build Your Foundations

   1. The mathworks: algebra, probability, reinforcement learning, decision trees… and more!
      📘 Try: Stanford CS229: Machine Learning (Resources)

2. Learn Core Machine Learning

   1. Grasp the core ML concepts that underpin modern GenAI.
      📘 Try: Andrew Ng Machine Learning Specialization (Resources)

3. Explore Generative AI Concepts

   1. Deep Dive into LLMs: ChatGPT, DALL·E, diffusion models, GANs, reinforcement learning.
      📘 Try: MIT 6.S087: Foundation Models & Generative AI (Resources)

🧭 For Non-Technical Leaders: Understand and Apply AI (Free)

1. Get AI-Literate Without Coding

   1. Learn the fundamentals of AI and how it applies to business: Covers: key terms, AI strategy, ethics, and implementation guidance.
      📘 Try: Andrew Ng AI For Everyone (Resources)

2. Explore AI in Business Contexts

   1. See how AI can drive transformation across industries: AI strategy, future of work, ChatGPT in enterprise.
      📘 Try: Wharton AI For Business Specialization (Resources)

3. Stay Updated Through Real Conversations

   1. Follow builders, researchers, and founders shaping AI’s real-world use.

      1. Listen to: Latent Space, No Priors, The AI Daily Brief

      2. Follow: karpathy, lex fridman,

🚀 Big Launches From Top Players

With most folks gone for vacation, I’d thought July would be a quite month… but quite the contrary - there were tons of amazing products released this month. Let’s take a look:

• [Infra] AWS AgentCore on Bedrock A managed runtime, memory, identity, browser, gateway & code‑interpreter stack for scaling any agent framework on Bedrock.

• [Apps] ChatGPT “Agent Mode” – rolled out to Plus/Pro/Team on 17 Jul 2025; the bot now spins up a virtual computer, browses, executes tools and finishes multi‑step tasks without hand‑holding.

• [Model] Kimi K2 – open‑weight 1‑trillion‑parameter Mixture‑of‑Experts LLM released 11 Jul 2025 by Moonshot AI; emphasises long‑context coding and autonomous tool use.

• [Model] Mistral Voxtral – dual‑size (24 B / 3 B) Apache‑2 speech model dropped 15 Jul 2025; combines ASR, semantic Q&A, summarisation, multilingual support and direct function‑calling.

• [Model] Alibaba Qwen3‑Coder – open‑source coding LLM unveiled 23 Jul 2025; 480 B‑param MoE variant excels at “agentic AI” code workflows and outperforms DeepSeek V‑series in lab tests.

• [Coding] Claude Sub‑Agents – Anthropic added specialised “crew” agents to Claude Code on 25 Jul 2025, letting multiple expert bots tackle architecture, tests, perf, docs in parallel

• [Coding] AWS Kiro CLI – free‑preview AI coding assistant launched 14 Jul 2025 (later throttled to cope with demand); integrates Claude Sonnet/Opus and soon Gemini 1.5.

• [Coding] Google Gemini CLI – 25 Jun 2025 preview (included for continuity): open‑source terminal agent built on Gemini 2.5 Pro; generous quota (60 req/min, 1 k/day).

💡 Startup Tips - User Interviews: We All Know What It Is, But Are We Doing Them Correctly?

By: @Jeffrey Hang

This is nothing new - if you’re building something new, you need to talk to people. Not just once, not just when something goes wrong, but consistently. The key is approaching it with the right mindset, strategy, and curiosity. We all know we need to do it, however it is very easy to get it wrong. Why? We might lose before we even start asking people.

Start by figuring out who you should be speaking with. It sounds obvious, but many entrepreneurs skip this step or go too broad. “Everyone” is not a target customer. Think about the people who truly have the problem you’re solving. Are they young professionals in cities? Are they overworked hospital administrators? Pet owners who travel frequently? Get specific. And don’t forget to identify your early adopters, they are the ones who feel the pain most acutely or love trying new things first. In B2B settings, also map out the decision-makers, blockers, users, and economic buyers. Knowing the internal dynamics can make or break your research and sales efforts later.

Once you know who to reach, the next step is understanding what you want to learn. It's easy to fall into the trap of asking people if they like your idea or would buy your product. But speculative answers like that are misleading. Instead, focus on uncovering real stories. Ask people to walk you through the last time they faced the problem your solution aims to fix. Listen for what they did, how they felt, what they tried, and what didn’t work. That’s where the gold is. Ask about the problem, not the product!

When it’s time to present a prototype or concept, resist the urge to jump in too early. First, let them talk about their needs and experiences. Otherwise, your shiny demo might bias the conversation. And when you do share a prototype, don’t expect brutally honest feedback unless you ask for it directly. Make it clear that you want the criticism because building something no one cares about is the real risk. It is rare for strangers to openly criticize an idea to your face after you just explained to them passionately.

Getting people to talk takes creativity. Referrals work wonders, so always ask each person to point you to two or three others. Meet people where they are, literally. Whether it's chatting outside a conference, in a waiting line, or even hosting an event, think about when your audience is most accessible and open to talking. If you're targeting enterprise buyers, use LinkedIn, call gatekeepers, or send concise emails asking for advice. People are more likely to respond if they’re not being pitched.

When you do land the interview, make it count. In-person is best, video is good, phone works in a pinch, written response…meh. AVOID focus groups, they dilute insights. One-on-one conversations give space for real stories to emerge. Having a note-taker helps, but if you’re solo, just make sure to write things down right after. And always start with a warm, human tone. Ask about where they’re from, their role, or their pets—whatever helps build comfort before diving in.

Stay aware of your own biases. It’s incredibly easy to hear what you want to hear. Go in prepared to be surprised, and even disappointed. You’ll learn more that way. Let them talk. Don’t rush to fill the silence. Don’t try to sell. Ask “why” more than once. Repeat things back to confirm what you heard. Sometimes just misphrasing what they said will prompt them to clarify their actual thoughts. We know our ideas are amazing, we want to know where it sucks!

When the interviews are done, the real work begins. Organize your notes, look for patterns, and share with your team. Sticky notes on a wall, spreadsheets, group debriefs, whatever works for you. Set goals in advance and measure your findings against them. But don’t get lost in the numbers. Use them to sharpen your instincts, not to chase false precision.

One of the strongest signals that you’re on to something is when people have tried to solve the problem themselves. Maybe they’ve cobbled together workarounds, or spent serious time and money fixing it. That’s a sign of real need. But don’t stop at identifying a problem, ask where it ranks in their priorities. Even painful problems don’t always make the top of the to-do list.

All of this effort, interviews, analysis, refinement, should point you toward better decisions. But it’s not a substitute for vision. You still need a belief about how the world could be better. Customer discovery helps you reality-check that vision and make sure you’re solving problems that matter.

And just because you’ve talked to 10 or 20 or even 50 people doesn’t mean you’re done. Keep talking. Shift your questions as your product evolves. Keep learning. Observation can be just as powerful as conversation—watch how people behave, not just what they say.

Building something people want isn’t magic. It’s the result of listening, testing, iterating, and staying curious. Start there, and you’re already ahead of most.

📖 Builder’s Corner - Agents? Automation? Let’s see how it handles news

The past few months we have seen an explosion in terms of agentic tool offerings, and these tools unlocks unparalleled possibilities. Equipping the agents with tools expands their action space, which in turns improves the problems they can reliably solve. ChatGPT agent mode is the hallmark of them all, acing all the benchmarks - quite literally.

So what?

Historically, you would need to rely on heavily enterprise focused products to optimize your workflows such as zapier - which is good, but expensive, inflexible, and actually difficult to onboard.

With the recent changes of AI, it brings about two changes

• Automating non-private workflows is getting easier & highly doable for non-tech folks

• Commercial grade products is getting amazingly powerful

These trends combined means that we are ‘inevitably’ entering an era where each and everyone of us can create ‘powerful’ apps that really just suits our needs. It ultimately comes down to a choice between build vs buy - and we have solid choices for both.

So… Any thing I can try out?

One challenge we might encounter is tons of news updates we get daily, or research updates… or any updates really. The challenge becomes how do we quickly filter through the content, dive deep where needed, while making we don’t drop a ball.

Here I’m going to show the simplest workflows possible - one that takes you ~5 minutes and gets you off the ground.

Quick & Dirty: RSS Feed + Google Sheet + ChatGPT

The concept is really simple here: gather all your news feeds (input), store them in a data sink (google sheet), and simply query them (chatgpt DR).

Connecting Your Data

There are many ways one can go about this. For starters, google sheet has this function `=IMPORTFEED("https://example.com/rss", "items", TRUE, 20)` you can paste into your cells. Be aware of the limits though - 50 feeds, refreshes hourly.

Now, if you need to have something that handles 20-100 feeds, can refresh every few minutes, then you can Make a Copy of our script, setup the triggers, and then add your feeds.

• Our RSS Feed Template

Search Your Data

Once you have the data ready, it becomes much easier to comb through them. Simply connect your data to ChatGPT (this is the only commercial one that works, today).

Once you got your data connected, you can simply schedule a task the summarizes the news/articles etc for you! (remember to add web search for the task). And while you are doing this, be sure to use O3 to do the summary and search. Combing through a google sheet is not a task for 4o.

Now, you can another option with agents - ChatGPT Agent mode. This gives it a bit more capabilities - such as pulling a lot more data, or follow specific instructions to interact with a website. This may sound pretty simple but keep in mind - A. ChatGPT Agent mode has only 40 usage for plus users (400 for pro). B. It cannot retain your credentials so anything requiring log in doesn’t work.

Finally, while we won’t cover it in this episode - but there are ways to replace the whole ChatGPT flow with customized local agent and MCP. You still need to pay for the API but the workflow will be more flexible.

The source code of the script, alongside our other toolings like Live Html Editor can be found at our github!

Have feedback or want to contribute? Reply to this email or join our community Discord. Until next time, keep building!

—Nick & the PMAI Team

💡 Editor's Note: NY Techweek was a whirlwind that left us with much to think about. During the very same week, AI Engineer’s world fair took place in SF (session recordings here). The session re-emphasized the concept of data-flywheel. A quite simple concept really - more user using your product, more datapoints you get to eval your product, the better you can make your product, which in turn attracts more users. It’s becomingly increasingly powerful as we enter a time where Agentic products are being created and shipped en masse. Compounded with more powerful models, better memory system and cheaper costs, we are going to see some amazing products that’s tailored ‘for you’ in the near future!

🎯 Key Insights

1. From Demo to Deployment

The shift is real - we're seeing fewer "wow, look at this!" demos and more "here's how we made it work at scale" discussions. Events like Bloomberg's production GenAI talk and the MLOps mini-summit are gold mines for practical insights.

2. The Memory Challenge

Multiple events focusing on AI agent memory systems? That's not a coincidence. As we build more sophisticated agents, memory becomes the differentiator between a toy and a tool. Don't miss the July 29th roundtable!

3. Infrastructure Matters

From Google Cloud Run GPUs to Kubernetes orchestration, July's events emphasize that great AI needs great infrastructure. The hackathon is a perfect opportunity to get hands-on with enterprise-grade tools.

🌟 Featured Events

• [Application] Pioneering Minds AI Workshop: Zero-Fluff MCP Sprint | Jul.11 | RSVP Join our friend Ben Morss from DeepL on a hands-on workshop that demystifies Model Context Protocol (MCP)!

• [DeepDive] AWS Summit NYC | Jul 16 | RSVP This complimentary event is your opportunity to learn from industry leaders, collaborate with peers, and get the answers to your questions directly from AWS experts. Join Swami Sivasubramanian, AWS VP of Agentic AI, as he reveals the services, capabilities, and techniques that are making it easier than ever to build and adopt AI agents.

• [Application] Agentic AI App Hackathon with Google Cloud Run GPUs | July 19-20 | RSVP Two-day hackathon to build and deploy agentic applications on Google Cloud, leveraging open models like Gemma, Llama 2, and Stable Diffusion XL on serverless NVIDIA L4 GPUs!

📅 Key Events This Month

This month's events reveal a crucial trend: we're collectively figuring out how to make AI systems that don't just demo well but actually survive in production. Here's your guide to leveling up from prototype to production!

🔧 IRL

• AI/ML Conversations Meetup: Production-Ready GenAI | July 8 | RSVP Bloomberg's Alex Tsankov shares realities of deploying customer-facing LLMs in data-rich enterprises. Perfect for understanding enterprise-scale challenges!

• NYC Agentics Foundation Meetup | July 10 | RSVP 45-60 minutes of rapid-fire demos on agentic AI, swarm architectures, and open-source agents. Pizza included - because debugging is better with carbs!

🤖 Application

• Building AI That Doesn't Break | July 2 | RSVP Virtual mini-summit for AI engineers and MLOps professionals. Learn about process calling for agentic tools, durable workflows, and orchestrating Kubernetes with Python. No fluff, just practical insights!

• Windsurf Brooklyn Build Session | July 8 | RSVP Guided building session with Windsurf for AI-powered development. Bring your laptop or use venue PCs (first come, first served). Food, drinks, and swag provided!

• Pioneering Minds AI Workshop: Zero-Fluff MCP Sprint | Jul.11 | RSVP Join our friend Ben Morss from DeepL on a hands-on workshop that demystifies Model Context Protocol (MCP)!

• Agents in Production 2025 | July 17 | RSVP Virtual event tackling the messy side of AI agents - memory management, behavior control, latency tradeoffs, and real deployment stories from the field.

• How it's Made: Creating a Multi-Agent Email Generation Tool | July 29 | RSVP Deep technical dive into production multi-agent systems, dynamic model selection, QA pipelines, and cost optimization strategies.

• Agentic AI App Hackathon with Google Cloud Run GPUs | July 19-20 | RSVP Two-day hackathon with mentoring, networking, and awards. Build with Gemma, Llama 2, and Stable Diffusion XL on Cloud Run's serverless GPUs!

🛠️ Deep Dive

• The Future of Neurotechnology: NYC Event | July 15 | RSVP Special evening with Dr. Ramses Alcaide from Neurable exploring brain-computer interface technology and its applications in mental health, productivity, and hands-free control.

• AWS Summit NYC | Jul 16 | RSVP This complimentary event is your opportunity to learn from industry leaders, collaborate with peers, and get the answers to your questions directly from AWS experts. Join Swami Sivasubramanian, AWS VP of Agentic AI, as he reveals the services, capabilities, and techniques that are making it easier than ever to build and adopt AI agents.

• Unlocking Memory in AI Agents | July 29 | RSVP Roundtable with top founders exploring tactical solutions for AI agent memory. Moderated by ex-LangChain engineer who developed LangGraph!

🔮 Final Thoughts

🎉 P.S. We got our own site!

Over the past month our volunteers have worked on building our very own platform pioneeringminds.ai for all of our events, speakers, resources, opportunities and more! In the coming month we will roll out features such as hiring recommendation, mentorship programs and AI courses!

Sign up today and lock in the perks! First 20 signups will receive FREE access to one of our in-person events!

So many exciting things happened in June - NY Techweek, AI Engineer’s World Fair, AI Startup School, Gemini CLI, Meta’s superintellience labs… Just to name a few.

With the development increasingly competent AI tools, the quantity of quality AI products will rise rapidly. With that comes a time where startups are forced to focus on the ‘productionization’ of their hacky startup projects and no doubt there will be an increased focus on production systems and reliability of the services. After all, no one likes using a product that ‘shuts down’ frequently.

With all that said, what excites me most? The combination of deep technical content (multi-agent architectures, memory systems) with hands-on building opportunities (hackathons, build sessions). Theory meets practice - exactly where innovation happens!

Don't try to attend everything. Pick events that align with your current challenges. Building agents? Focus on the agentic AI events. Scaling to production? The MLOps and Bloomberg talks are your jam. Just exploring? The hackathon and build sessions offer low-pressure learning.

See you out there, builders! And remember - in production, no one can hear your demo scream. :D

📚 References & Reading List

• Anthropic (2025). How we built our multi-agent research system. https://www.anthropic.com/engineering/built-multi-agent-research-system

• Cognition (2025). Don’t Build Multi-Agents. https://cognition.ai/blog/dont-build-multi-agents#a-theory-of-building-long-running-agents

💡 Editor's Note: We’ve had an incredible NY Tech Week! We’re so grateful to everyone who made it unforgettable. Over the past weeks, we hosted three amazing events. We kicked things off with a startup panel, then our Hack-for-Impact hackathon, and wrapped up the week with a vibrant demo day. Thank you to everyone who joined us, supported the community, and especially to our generous sponsors who helped bring it all to life. We had over 1000 sign-ups this tech week and we cannot wait to continue building NY’s AI world.

🌟 Coming Up Next!

Knowledge sharing doesn’t take a break - our line ups won’t stop either!

• [Application] Pioneering Minds AI Workshop: Deep Research Agents with Google ADK | Jun.27 | RSVP Join our mysterious speaker on building your own ‘deep research’ agent in an hour!

• [Application] Pioneering Minds AI Workshop: Zero-Fluff MCP Sprint | Jul.11 (TBD) | RSVP Join our friend Ben Morss from DeepL on a hands-on workshop that demystifies Model Context Protocol (MCP)!

From Partner Communities!

• [Deep Dive] Smarter ML: Infra, Transformers & Optimization Hacks | Jun.23 | RSVP Join NYC ML Ops for an evening of real talk focused on infrastructure scaling, rethinking attention, and cost-efficient pipeline design.

• [IRL] AI Hot 100 Summit | Jun.26 | RSVP 100 of the Hottest AI Startups in the World – the Who’s Who of Early Stage AI, All in One Place.

  • Our members gets 10% discount. Please reach out for discounted ticket!

🔍 Featured Resources

Majority of our sessions are recorded. Please find our past sessions, articles and books here.

💼 Hiring Board

Our community partners are actively recruiting top talent! Check out these exciting opportunities:

• LastMile AI

  • Product Growth Lead, Senior ML Engineer, Staff Backend Engineer & Staff Fullstack Engineer

• CoGrader - Founding Sales

• Nunchi - Engineering Lead

• DeepL - Multiple positions across engineering, product, and more

• Have a job to share? Let us know by replying to this email!

🛠️ For AI Builders

Take advantage of these exclusive resources from our partners:

• NachoNacho Deal Hub - Access exclusive discounts on SaaS tools for our community

• LastMile AI MCP Agent - Open-source Model Context Protocol agent implementation

• DeepL Free API - Get started with DeepL's powerful translation API at no cost

🎙️For AI Founders

If you startup is looking for fund-raising (or revenue-sharing)

• Agent Boostr supports AI startups with proven ARR Priscilla Liu!🎉

• Taihill Venture supports early stage AI startups!

🌟A16Z New York Techweek Recap!

🎉 AI Founders Panel: The Good, The Bad, The Unknown

We were honored to host an inspiring panel with Andrew Hoh (LastMile AI), Gil Quadros Flores (CoGrader), and Trista Van Tine (Nunchi), who shared honest, behind-the-scenes perspectives on building in AI: from navigating rapid market shifts to staying grounded through the highs and lows of startup life. The conversation covered everything from product-market fit and technical hurdles to ethical challenges and what it means to lead in a field where the rules are still being written. More Info here.

🌟 Hack For Impact NYC - June 6, 2025

The Hackathon was an intensive one-day sprint to build AI-powered solutions under active mentorship from industry experts at Google, Meta, and Microsoft, DeepL, and many more! With participants working in teams across four mission-driven tracks, Education, Climate, Urban Challenges, and Responsible & Ethical AI, the event offered a hands-on opportunity to create end-to-end products and compete for cash prizes totaling $1,100 and a slot to pitch at the Builder’s Demo. Huge thank you to our sponsors Polsinelli, Lastmile AI, and 1 Thing Org. More info here.

💥 Builder's Showcase - June 7, 2025

We finished our week with our Showcase that brought together talented AI innovators to demo their latest breakthroughs, including Lion Lease, See Museums, SuperRent AI, Formalyze, Train4AI, and the winners from Hack4Impact 2025 (Parents In the Loop)! The session focused on knowledge sharing, inspiration, and connections among builders in the rapidly evolving AI space, offered by the interactive demos, Q&A, and networking, giving attendees direct exposure to real-world AI products and use cases. More info here.

🎉 In Case You Missed It

How to build your first AI-Driven App with Jing Dong @ Microsoft!

(Recording & Deck)

​We were incredibly fortunate to have Jing Dong, a ML scientist at Microsoft, to host a workshop for us. She broke down the intimidating stuff into bite-sized steps from her experience in shipping AI solutions at Microsoft while building indie apps on the side. She shared her struggles of translating ideas into code. If you’ve sketched an AI idea on a napkin but never shipped a working product, this one-hour “vibe-coding” sprint was an unmissable event.

A big shoutout to our Techweek sponsors!

Polsinelli Law Firm is an Am Law 100 firm with more than 1,000 attorneys in 22 offices nationwide. The firm's attorneys focus on health care, finance, real estate, technology, private equity and life sciences.

1 Thing.org a grassroots, non-profit organization of volunteers dedicated to fighting anti-Asian racism.

DeepL isn't just a translator—we're a comprehensive Language AI platform that enables organizations to communicate effectively across languages, cultures, and markets.

Heynunchi.com is an AI-powered platform that helps you understand, connect, and communicate with your members in ways that weren't possible before. Turn data into relationships, insights into actions, and members into advocates.

lastmileai.dev is an enterprise-grade evaluation platform that provides the essential tools to enable developers to test, evaluate, and benchmark AI applications.

Taihill Venture is a Boston-based, industry-agnostic pre-seed deeptech fund, dedicated to pioneering breakthrough technologies that tackle longstanding challenges.

💡 Editor's Note: Hope everyone had a great holiday! Can’t believe June is almost around the corner. This Summer is a bit chiller than usual but, hey, we’ve got the events to heat us up! Better yet, NY Techweek is around the corner. Definitely check it out if you want to learn about what’s going on in the tech space!

As someone who's been tracking the AI ecosystem closely & using it quite a lot… (remember my 5.5 million tokens to o1pro in March? 😅), I can tell you that this month's lineup is just, AWESOME. Whether you're looking to connect with fellow founders, dive deep into AI agents, or learn about the landscape of AI, June has something special for everyone!

🌟 Featured Events

• [IRL] TechPulse 2030: AI Revolution | June 5 | RSVP An immersive evening exploring how responsible AI is transforming financial services, healthcare, and impact sectors. Join visionary leaders from AI 2030, Lair East Labs, AWS, and FinTech4Good for bold conversations that inspire action!

• [Application] Open-Source NYC Computer Vision Hackathon | June 21-22 | RSVP Two-day hackathon using Raspberry Pi, camera modules, and Moondream's vision model to develop 'Edge Vision × NYC Open Data' solutions for civic impact.

• [Deep Dive] Deep Dive into Deep Tech Startups | June 4 | RSVP Hear from three NYC-based founders (Arena, Albedo & Fauna Robotics) sharing insights on starting and joining a deep tech startup!

Hackathon Workshop

If you are going to our Hack For Impact 2025 hackathon, be sure to sign up for this workshop to learn about how to get your ideas out blazing fast!

• [Application] How to build your first AI-Driven App | June 1 | RSVP The workshop by Jing from @Microsoft covers the steps to vibe-code a fully functional app with frontend, backend, Google-auth configured!

🚀 NY Tech Week Special: Where Founders Build Together

Want to meet people building cool things and develop a killer network of founders? I've curated the must-see events for builders and aspiring founders!

🔨 Build & Ship Events

• Work Smarter, Not Harder: Build an AI Agent in 5 Minutes | June 3 | RSVP Build your own AI agent using IBM's watsonx Orchestrate - no experience needed! Perfect for meeting other builders in a hands-on environment.

• FoundersNetwork AI OpsLab | June 4 | RSVP Dive into AI-powered Go-To-Market tech stacks with founders from ScaleStack.ai, Sellible.ai, and Bounti.ai. Real-world GTM use cases you can implement immediately!

• Google Build with AI NYC | June 5 | RSVP Intensive workshop building production-ready AI agents with Gemini 2.5. Limited spots, bring your laptop and government ID!

💡 Founder-Focused Panels & Networking

• How to Build an AI Startup that Attracts Funding + Customers + Talent | June 4 | RSVP George Bandarian from Untapped Ventures walks through frameworks for building venture-backable AI startups. Perfect for technical builders transitioning to founders!

• Navigating the Future of Agentic Development | June 4 | RSVP Panel with Erik Nordlander (Google Ventures), leaders from Datadog, Graphite, and Warp discussing how AI agents are reshaping software development.

• Roundtable: Building Emotional Support Systems with AI | June 5 | RSVP Women-led intimate roundtable hosted by EchoHer and PowerYou exploring how to build AI for emotional wellbeing, digital wellness, and addiction recovery. Break into small groups with mission-driven founders to discuss designing compassionate AI that truly serves users!

🤝 Elite Networking Opportunities

• NYC Founders' Fireside: Exit Stories | June 5 | RSVP Learn from founders who've achieved multi-million dollar exits. Cocktail reception included for extended networking!

• a16z speedrun x Adobe: Where AI Meets Creativity | June 5 | RSVP Fireside chat with Andrew Chen (a16z) and Scott Belsky (A24/Adobe). Due to capacity constraints, RSVP early - this is where the top builders gather!

Pro Tip: The best connections happen at the smaller, more focused events. Don't just hit the big panels - join the tight-knit workshops!

📅 All June Events

🎉 June 3-4

IRL

• AI Agents & Co-Pilots Panel: Shaping the Future of Business | June 3 | RSVP

• AI/ML Conversations Meetup: When AI Beats Physics | June 3 | RSVP

• Fullstack Infra for AI: Building Type-Safe Systems for Vibe Coders | June 3 | RSVP

• MCP and the future of AI x API | June 4 | RSVP

Application

• Tech for Good Forum: Bridging AI Innovation and Scalable Impact | June 3 | RSVP

• AI Agents in the Enterprise | June 3 | RSVP

• Work Smarter, Not Harder: Build an AI Agent in 5 Minutes | June 3 | RSVP

• How to Build an AI Startup that Attracts Funding + Customers + Talent | June 4 | RSVP

• Innovate for People & Planet : Fintech & AI | June 4 | RSVP

Deep Dive

• Vibe Coaching with AI VC George Bandarian @ Untapped Ventures | June 3 | RSVP

• Building and Securing GenAI Apps | June 3 | RSVP

• AI in-action: Driving REAL Impact | June 4 | RSVP

• "The ROI of clean code" with JPMC and LSEG | June 4 | RSVP

• Infra: The New Moat in AI | June 4 | RSVP

• Deep Dive into Deep Tech Startups | June 4 | RSVP

• Navigating the Future of Agentic Development | June 4 | RSVP

🎉 June 5-7

IRL

• NYC Founders' Fireside: Exit Stories | June 5 | RSVP

• TechPulse 2030: AI Revolution | June 5 | RSVP

• AI Search & SEO: The New Era | June 5 | RSVP

Application

• Roundtable: Building Emotional Support Systems with AI | June 5 | RSVP

• a16z speedrun x Adobe: Where AI Meets Creativity | June 5 | RSVP

• Storytelling & AI In The Age Of Growth Marketing | June 5 | RSVP

• AI and the Future of Government (Tech) | June 5 | RSVP

• Faith in AI | June 5 | RSVP

• New York's AI Edge: Leading with a Future-Ready Workforce | June 6 | RSVP

• Business Impact with Open-Source AI | June 6 | RSVP

Deep Dive

• FoundersNetwork AI OpsLab: The Ultimate AI-Fueled Go-To-Market Tech Stack for Startups | June 4 | RSVP

• Google Build with AI NYC | June 5 | RSVP

🎉 Through June

• GenAi in Games, 3D Animation and VFX | June 10 | RSVP

🎯 Strategic Event Planning Guide

For Aspiring Founders:

1. Start with "How to Build an AI Startup" (June 4) for frameworks

2. Join "Work Smarter, Not Harder" (June 3) to build your first agent

3. Network at "NYC Founders' Fireside" (June 5) for inspiration

For Technical Builders:

1. "Google Build with AI NYC" (June 5) for hands-on experience

2. "Deep Dive into Deep Tech Startups" (June 4) for technical insights

For Investors & Business Leaders:

1. "AI Agents in the Enterprise" (June 3) for implementation strategies

2. "Business Impact with Open-Source AI" (June 6) for regulatory insights

3. "Infra: The New Moat in AI" (June 4) for investment perspectives

🔮 Final Thoughts

This Tech Week is shaping up to be something special. The curated events focus on ‘building and shipping AI products’ at the frontier of respective fields. There are many other awesome Techweek AI events, so make sure to check out the calendar and get on the waitlist!

My advice? Don't try to attend everything. Pick 3-4 events that align with your goals (Ask: why am I attending this?) and go deep. The real value comes from the conversations after the panels, the debugging sessions during workshops, and the "aha!" moments when you're building alongside others.

I'll be at several of these events - if you see me, please say hi! I love hearing about what you're building. :D

Happy building, and see you at Tech Week!

💡 Editor's Note: Following up on our previous post, here are the promised New York Techweek event line ups!

📢 Call To Action!

Do you have a product to show? Apply today to join the NY Techweek AI Builder's Showcase! What you built will be shown to the community as well as investors! Limited slots available and we will accept on a rolling basis!

🌟A16Z New York Techweek Lineup!

🎉 AI Founders Panel: The Good, The Bad, The Unknown [RSVP]

• Date: June 4th, 17:00-19:00 Wed, 2025

• What to Expect: A candid discussion with founders who have successfully raised funding and have deep industry experience. They'll share their journeys navigating the unprecedented AI revolution happening right now - both the extraordinary opportunities and distinctive challenges.

• Confirmed Panelists

  • Trista Van Tine, Founder, Nunchi AI, Michigan Tech Week

  • Andrew Hoh, Co-Founder, LastMile AI

  • More to come!

🌟Deliver Innovative Solutions For Societal Impact: Hack For Impact NYC 2025 [RSVP]

• Date: June 6th, Fri, 09:00-19:00 2025

• What It's About: A one-day initiative designed to channel the power of Agentic AI into tangible solutions for NYC's most pressing needs. Taking place during NYC Techweek, this event brings together diverse teams—students, engineers, and civic leaders—to tackle urban challenges head-on.

Recommended Tracks

• A. Education Equity

  • Tools for remote learning, language support for immigrant families, or bridging the digital divide in underserved communities.

• B. Climate Resilience and Sustainability

  • Solutions to manage waste more effectively, reduce carbon footprints, or adapt infrastructure for extreme weather events.

• C. Urban Challenges

  • From famed rodent challenges to housing affordability, from hate crime monitoring to mental health support, there are vast amount opportunities to leverage cutting edge AI to solve our toughest problems at hand.

• D. Responsible and Ethical AI

  • Frameworks for reducing AI bias, ensuring transparency in AI decision-making, protecting privacy in data collection, and creating accessible AI systems that work equitably for diverse populations including those with disabilities or from marginalized communities.

Top 3 Teams will be automatically granted a slot in our Builder’s Showcase!

💥 Show your work! AI Builder's Showcase [RSVP]

• Date: June 7th, Sat, TBD, 2025

• What It's About: The perfect opportunity to see innovative AI projects from our community, including Hackathon participants! Connect with like-minded builders and find inspiration for your own projects.

• Agenda: This session will provide opportunities for 10 teams to show case their products (5 mins each), and get live-feedback from audience (5 mins)!

NOTE: We promote and accept functional products for live-demo only. If your product is still in ideation stage, we encourage you to join our Hackathon!

🔍 Relay App Workshop Recap

Now, a big thanks to everyone who joined our workshop with Jacob (jacob@relay.app) last Friday! It was fantastic seeing you all build LinkedIn profile finders in real-time.

My favorite moment was when Jacob demonstrated how breaking complex tasks into micro-steps makes AI much more accurate. Admittedly, finding my profile with mismatching information was a tougher challenge than expected! :D

As promised, here's everything you need to continue building with Relay:

• 📹 Workshop Recording

• 📝 Clone the LinkedIn Finder Template

• 📺 Relay YouTube Channel for more tutorials

Note: Can't access the recording? Just subscribe

Subscribe now

and you'll gain access automatically! (The subscription trigger was built with Relay and took me literally 10 seconds to set up - that's how powerful these workflow tools are!)

Have you built something cool with what you learned? We'd love to see it! Tag our Linkedin account when you do!

🎉 Congratulations to Our Manus Winners!

A big round of applause to Ow**n, Ok**y, and Jere**h for winning access to the Manus Community Account! Your comments and enthusiasm for AI made you stand out in our recent giveaway. You'll be receiving access details directly via email.

Didn't win? Don't worry - we'll have more opportunities coming soon as we continue building relationships with cutting-edge AI tool providers.

🌟 Final Thoughts

The next two months represent an exciting opportunity to deepen your AI skills, expand your network, and possibly even launch your next big project. What makes our community special is the combination of learning and doing - theoretical knowledge put immediately into practice.

Whether you're just getting started with AI or already building sophisticated applications, there's something for you in our upcoming events. I'm particularly excited about the Hack for Impact NYC event because it demonstrates how quickly AI tools are evolving from novelties into instruments of real social change.

Will I see you at one of our upcoming events? I hope so! And if you have ideas for future events or would like to contribute in any way, please reach out - this community grows through your participation.

Happy building!

A big shoutout to our Techweek sponsors!

Polsinelli Law Firm is an Am Law 100 firm with more than 1,000 attorneys in 22 offices nationwide. The firm's attorneys focus on health care, finance, real estate, technology, private equity and life sciences.

Heynunchi.com is an AI-powered platform that helps you understand, connect, and communicate with your members in ways that weren't possible before. Turn data into relationships, insights into actions, and members into advocates.

lastmileai.dev is an enterprise-grade evaluation platform that provides the essential tools to enable developers to test, evaluate, and benchmark AI applications.

And a huge thanks to our community sponsors!

Relay.app is a platform to create AI Agents that work for you across Gmail, Notion, HubSpot and 100+ other apps.

sanctum.ai At Sanctum, our vision for the future of AI places user privacy squarely at the center.

Superlinear Academy offers immersive, project-based learning experience that transforms passive AI users into confident builders.

NachoNacho is the world's largest marketplace for SaaS and professional services. Our mission is to harmonize the subscription and services economies and foster their further growth. We use the latest fintech and AI tools to create powerful user experiences for all stakeholders in the marketplace.

References & Reading List

1. Identifying and scaling AI use cases, OpenAI, 2025. https://cdn.openai.com/business-guides-and-resources/identifying-and-scaling-ai-use-cases.pdf

💡 Editor's Note Can you believe it's already May? This month, we're seeing an incredible convergence of AI governance, practical applications, and forward-thinking events that are shaping the industry landscape. I've been particularly excited about the emerging focus on AI agents and their real-world applications (more on that below!).

This month, we're also highlighting upcoming AI conferences to help you plan the AI events through 2025! These events offer excellent opportunities to connect with industry leaders, network with fellow AI enthusiasts, as well as learning about the forefront of AI! Check out our conferences section to plan your calendar!

Sincerely sorry for the mistakes in the previous post. Links have been fixed and we have added two additional events.

🌟 Featured Events

• [IRL] Founder's Panel with Jerry Limber: Navigating the Entrepreneurial Journey | May 2 | RSVP Join PenguinHub.ai co-founder Jerry Limber for an interactive panel offering strategic insights on scaling ventures, practical founder advice, and balancing innovation with operations!

Bonus event for community! If you joined our events before, this one is on us!

• [DeepDive] Workshop on Emerging Trends in AI | May 5-6 | RSVP This two-day workshop brings together leading experts in machine learning and neuroscience to examine resilient systems and ML-driven data generation in social sciences.

• [Application] Generative AI and Human Capabilities | May 8 | RSVP Nicolas Chaillan shares his journey of creating a company using generative AI, taking his vision to $1M in sales in under 18 months.

📅 Key Events This Month

IRL

• NachoTuesday: SaaS & AI Founder + Investor Happy Hour | Apr 29 | RSVP Exclusive NYC rooftop networking event for SaaS founders, C-suites, investors, and tech leaders with sponsored drinks and appetizers.

• Founder's Panel with Jerry Limber: Navigating the Entrepreneurial Journey | May 2 | RSVP Join PenguinHub.ai co-founder Jerry Limber for an interactive panel offering strategic insights on scaling ventures, practical founder advice, and balancing innovation with operations!

• The OOO Summit 2025 | May 16 | RSVP ​The flagship event for entrepreneurs, creators, and ambitious builders, brought you to by Carry and Andrew Yeung.

  ​OOO stands for Owners, Operators, and Outliers. It is the ultimate gathering for entrepreneurs, operators, and disruptors looking to take their businesses, careers, and networks to the next level!

Sign up for OOO SUMMIT with PIONEERAI to get $100 OFF!

Application

• Intelligent Engagement: AI's Role in Sports and Entertainment | May 5 | RSVP Panel discussion on how AI is transforming engagement in sports and entertainment, from streaming to stadium experiences.

• Generative AI and Human Capabilities | May 8 | RSVP Nicolas Chaillan shares his journey of creating a company using generative AI, taking his vision to $1M in sales in under 18 months.

• Defining the Future of Public Health Preparedness in NYC | May 13 | RSVP Full-day event bringing together researchers, government officials, and innovators to explore cutting-edge technology solutions for public health preparedness.

• AI Agent Builders Summit: World Tour Kickoff | May 28 | RSVP Join the Bay Area kickoff event featuring 20+ practitioner booths, lightning talks, and keynotes from leaders building and productionizing AI agents.

Deep Dive

• The Role of AI Audits in Responsible AI | Apr 30 | RSVP Learn how AI audits help organizations build trust, minimize risk, and ensure AI systems are safe, explainable, and fair.

• GenAI Collective NYC: April Research Roundtable | Apr 30 | RSVP Research-focused event with Hebbia AI featuring talks from researchers at Morgan Stanley and beyond.BigAI Talk: Hybrid Planning-Learning Systems | May 2 | RSVP Caelan Garrett of NVIDIA discusses integrating task and motion planning with learning for robot data generation and deployment.

• AI + Open Education Initiative: AI Literacies and Evaluation | May 12 | RSVP Authors discuss papers on professional education, judicious use of AI, and open practices in this MIT Open Learning webinar.

• The Future of Design and Technology | May 20 | RSVP Engaging panel with Cornell's Department of Design Tech exploring how emerging technologies are transforming design practices.

🎟️ Upcoming AI Conferences

Now's the perfect time to plan your conference schedule for the year! Here are the most anticipated AI conferences of 2025:

Summer 2025

• Generative AI Summit | New York | June 5 | Register
  $100 off early bird pricing
  Bringing together engineers, developers, and executives at the forefront of generative AI revolution, with access to three co-located summits with a single pass.

• AI & Big Data Expo North America | June 5-6 | Register
  Free Expo Pass available
  Premier event showcasing the latest in Generative AI, Enterprise AI, and Machine Learning with 250+ speakers across multiple tracks.

• Data + AI Summit 2025 | June 9-12 | Register
  $1,295 early-bird until April 30, free virtual option
  The world's largest event for the data and AI community featuring 700+ sessions and 20,000 in-person attendees at Moscone Center, San Francisco.

• CDAO Government 2025 | June 25-26 | Register
  Free for government employees
  Premier event for senior data and analytics leaders across U.S. government agencies, focusing on enhancing transparency and improving citizen services.

• The AI Furnace Summit | Jun 26 | RSVP This exclusive, invite-only gathering brings together 100 of the hottest AI startups, Fortune 500 executives, and top investors in an intimate setting designed for meaningful connections. With active communities in London, Boston, and New York, The AI Furnace is building a powerful global network of AI innovators!

Fall 2025

• Ai4 2025 | August 11-13 | Register
  $1,695 early registration (47% discount)
  North America's largest AI industry event, bringing together business leaders and data practitioners at MGM Grand, Las Vegas.

• The AI Conference 2025 | September 17-18 | Register
  Super Early Bird pricing from $750
  Groundbreaking two-day event with 100+ visionary speakers exploring AGI, LLMs, and neural architectures at Pier 48, San Francisco.

• Coalesce 2025 | October 13-16 | Register
  $695 Early Bird until May 26, free virtual option
  Premier analytics engineering conference for data practitioners at Resorts World Las Vegas.

• QCon AI New York 2025 | December 16-17 | Register
  $990 Early Bird until May 13 (31% off)
  Focused on practical, real-world AI implementation for senior software developers and engineering leaders.

Pro Tip: Early bird pricing can save you up to 47% on conference registration! Most early bird deadlines are approaching quickly in April and May.

📋 Conference Planning Guide

Ok, so these are tons of information and conferences are a really big commitment. So, let me help you with some tips to plans this easier : D

Tips

1. Aim for virtual experience. Networking with folks is great. However, adding lodging, flight, car rental, and tickets… that’s huge. You can learn a lot from online sessions as well (and of course summaries after the event)!

2. Set a target for networking and listen. If you are an introvert like me, talking to people takes lots of ‘internal’ effort and energy. However, if I set a bar (say talk to 3 people is great enough), then it becomes much easier to go out there and just listen and connect. (Same trick applies to asking questions too. I am well aware of the sweat in my palms when I am working up the ‘guts’ to ask questions. But hey, if asking one question is good enough, I’ll just do it : D)

3. Check the Agenda. Make sure you know what’s going on - when and where and who. If a summit doesn’t have agenda… just wait until it does before signing up.

Guides

1. For Enterprise AI Leaders: Ai4 2025 and CDAO Government 2025 offer executive-focused content and networking.

2. For Technical Practitioners: Data + AI Summit and QCon AI provide deep technical content from leading practitioners.

3. For Analytics Teams: Coalesce 2025 is specifically designed for analytics engineers and data professionals.

4. For Startups & Entrepreneurs: The AI Conference and Generative AI Summit showcase cutting-edge innovations and investor connections.

Are you planning to attend any of these conferences? Share your plans with the community and organize a meetup!

🔮 Final Thoughts

May 2025 is, again, an action-packed month with events spanning governance, implementation, and forward-looking research. The increasing focus on AI agents and practical applications shows how quickly the field is maturing from theoretical concepts to real-world solutions. We are now talking evaluations, not just what AI agent can do.

Looking ahead to summer, the conference lineup offers something for everyone in the AI community - from technical practitioners to business leaders. These gatherings offer great insight for understanding the rapidly evolving AI landscape and building the connections that drive innovation.

Stay tuned for our next newsletter! We will be announcing our New York techweek events!

📚 References & Reading List

• AI Accelerator Institute. (2025). Generative AI Summit | New York. https://world.aiacceleratorinstitute.com/location/newyork/

• Corinium Global Intelligence. (2025). CDAO Government 2025. https://cdao-gov.coriniumintelligence.com/

• Databricks. (2025). Data + AI Summit 2025. https://www.databricks.com/dataaisummit

• Fora Group. (2025). Ai4 2025. https://ai4.io/vegas/

• MLOps Community. (2025). AI Agent Builders Summit: World Tour Kickoff. https://home.mlops.community/home/events/agents-world-tour-kickoff

• Northeastern University. (2025). Navigating AI Governance: What Leaders Need to Know. https://ai.northeastern.edu/event/navigating-ai-governance-what-leaders-need-to-know-with-dr-cansu-canca

💡 Editor’s Note: Last week when chatting with community members, I came to realize that we had a common need - summarize AI industry news and get a daily digest. Sounds like a daunting project - until I came to realize the AI-powered workflows can solve this nicely! I’ll detail the entire flow in the builders corner for your to build your own workflow!

💡 Share & Try Manus.im! Last week we had a great chat Manus founder Tao and got access to a few activation codes. To make sure everyone can try it out, we are providing a Community Manus Account for you to try out manus!

From now through April.18, share this post, comment your AI experience below, and we draft 3 lucky winners for access to the Community Account!

Join us to learn about AI Workflow (relay.app) April.18!

Sign up today at https://lu.ma/u9oblvlk !

On April 18, Jacob Bank, founder of relay.app, ex-Google, serial entrepreneur , will share how to create powerful Automation workflows leveraging AI agents. If you have heard about workflow tools like Zapier, Make, n8n etc. This is the new, up-n-coming tool that provides solid experience!

Join AI Founders/Aspiring Founders social on April.11!

Sign up today at https://lu.ma/dfcgpy9t !

​Looking to connect with fellow AI founders, exchange ideas, or just enjoy conversations about the latest in AI over drinks? Come hang out with us at Market 57 for an evening of casual networking and authentic discussions!

@Market 57 Next to Chelsea Market, we will have a light-hearted social gathering at the pier! **food not included**

Builder’s Corner - Get Creative With Workflows

Acharya et al (2025) defined Agentic AI as “refers to autonomous systems designed to pursue complex goals with minimal human intervention”. This is pretty much all the imagination we have about Agentic AI - a tool that just does the job for us, handsoff. However, such is not the case just yet for all of the tools out there.

Agentic AI and AI Agents covers a wide range of products with different levels of autonomy. On one end of the spectrum, there are tools like Cursor (agent mode), Manus (general AI agent), OpenManus (open source, coder-oriented version of Manus) that can perform complex tasks on their own and actually modify the environment. On the other end, there are tools like OpenAI Deep Research that can perform complex searches and analysis (yeah it can write & debug it’s code, try it out). Finally, there are tools that is a mixture of human-loops and agents - one that harness both the Agent autonomy and reliability.

This time we will focus on the AI workflow with a quick showcase of Full-autonomous agent like Manus.

1. Build AI-Drive News Summary (relay)

There are so many things happening since since the launch of ChatGPT (and even before that : D) and I find it difficult to stay up to date with all the information.

Don’t get me wrong, there are tons of newsletter services out there - Morning Brew, AlphaSignal, The Rundown etc. However, those ain’t what I needed.

I follow folks on twitter, I subscribe to youtube channels, economy has a deep impact on AI efforts, new policies in EU, my financial times articles… Most of these are not quite covered in these newsletters. Beside, I want to see the trend, the top stories, and only check it out if it catches my eye.

In other words, my requirements are

• See latest developments in AI/Finance/US from news, twitter, youtube and my FT channels.

• Get a short, succinct summary of daily news.

• Can check out articles in detail if I want.

Quite simple, right? However, it’s not so simple to implement from scratch.

1.1 The Cursor Way

Now, the native thinking to implement this is really straight forward - just deal with my requirements one by one. How?

Pull the news - I got inoreader access ($8/month for API access), I got the feeds I want. Then I do Oauth2, get refresh tokens, access tokens, store them and schedule a job to pull the news periodically (could be done locally, or cloud tasks, or AppScript, or… many other tools).

Summarize the news - this step is really simple. I write and tune my prompt, send all the news and what I want to the model, get structured response and then parse it. Gemini has large context window so it’s easy to do without any optimization to save contexts.

Send the updates. Surprisingly not so simple - I set up my flask sever, needed to configure my smtp mail server, set up my Google App passwords, format my contents in markdown, create pdf then send.

Don’t get me wrong, this is totally doable and I have done it. With the help of cursor it took me a large chunk of time, about 4 hours end to end.

And yes, cursor (agent) did the bulk of coding for me. Configuring inoreader oauth and combine it with google auth on supabase was a mess though.

That being said, if you don’t know what RSS feed is, this is another barrier right from the start.

An RSS (Really Simple Syndication) feed is a web feed format that allows users to access updates from websites in a standardized, computer-readable format, enabling easy aggregation and consumption of content.

— wikipedia

1.2 The Relay Way

Now, after playing around with the workflow mentioned above. I found one more issue - I need to either runs this locally, or migrate the workflow to AppScript. I could have done it, though at this point I’m out of patience hh. (Turns out to be good thing).

Quite frankly I had no idea what Relay was and has never used Zapier or the like before. I just figured that I need a tool to automate the workflow for me.

This is the entire flow - one that took me ~30 minutes (bulk of time on playing around with format).

In short, this does the same tasks I have mentioned above.

Pull the news - Schedule a run (daily at 6AM), refresh & store tokens (http request), pull the news (http request).

Summarize the news - this step is really simple 2. I bundled all the news I pulled, with some prompts for summary and and instructions for a json output. Gemini 2.5 pro does the job nicely. Most importantly, Gemini API is Free.

Send the updates. Again, super simple now. I literally spend 1 minute add a note to create pdf, 1 minute to create a node to send PDF to my inbox.

Done.

The result?

Sorry about the plain, industrial style summary… Though fancy UI / flashy color could be easily added / configured. See the result here. If you have no access, simply subscribe and you will gain access within an hour.

2. Do the job for me - AI

Now this has unlocked quite a few workflows that I have been thinking of doing for a while.

For example, to keep up with the latest publications across channels, I frequently read journals from Arxiv.org, PubMed, IEEE etc, lots of sites. Quite frankly, the load is overwhelming - especially at the current rate of publication/submissions. Now… all of these support RSS feed - you know what this means : D

I wanted a way to keep up with all my emails - basically a simple CRM and a reminder of what I do. How can this be done? Let’s take a look:

• Collect Emails

• Summarize Emails to Actions

• Send to my inbox / Schedule Calendar events

Each of these steps can be easily done with AI automation tool like relay (and zapier etc).

These tasks are easy when you can map them out - and find tools to just do them simply.

Summarize emails - this step is easily done with AI. How about the events? Well, turns out you can as easily ask gemini to generate required data for scheduling events as well as summaries. Once you have the data, this is done. In Relay (and Zapier, n8n, make etc), you can easily create a human in the loop interaction where you need to confirm the data for next steps. This is a good way to figure out what to schedule & reject the incorrect ones.

Fully autonomous Agentic AI tools is a topic for another day… However, just as we have covered in the last post, all the tools are just not there yet. Some are better, some are worse (manus is on the better end of the spectrum : D).

Final Thoughts - The Future of Workflow Automation

The real power of AI today isn't just in what a single model can do - it's in how we can connect these tools together to solve real problems in our daily lives. As we've seen, workflow automation tools like Relay allow us to bypass the complexities of coding from scratch while still creating powerful, personalized solutions.

A natural next step of this, is asking AI to create the workflow for us. Microsoft (2025) already has an article for this process, and just as the have said AI Agent Workflow creates a more deterministic behavior for the AI workflows that is much more suitable when we want reliability.

Think about it - what repetitive tasks are eating up your time? News summaries? Email organization? Content creation? There's likely a workflow waiting to be built!

💡 Key Takeaways:

• AI workflow automation bridges the gap between fully autonomous agents and manual processes

• You don't need to be a developer to create powerful AI-driven workflows

• Start small with a specific pain point, then expand as you get comfortable

I'd love to see what you build! Share your workflows or drop a comment below with your challenges - Let’s solve it together!

References & Reading List

1. Acharya, D. B., Kuppan, K., & Divya, B. (2025). Agentic AI: Autonomous Intelligence for Complex Goals – A Comprehensive survey. IEEE Access, 1. https://doi.org/10.1109/access.2025.3532853

2. Google (2024). What is an AI agent?. https://cloud.google.com/discover/what-are-ai-agents

3. Github (2025). What are AI Agents? https://github.com/resources/articles/ai/what-are-ai-agents

4. Microsoft (2025). Introducing agent flows: Transforming automation with AI-first workflows. https://www.microsoft.com/en-us/microsoft-copilot/blog/copilot-studio/introducing-agent-flows-transforming-automation-with-ai-first-workflows/

A Special Thanks To Our Friends!

Sponsors

• Sanctum.AI, Private Local AI Solution, Contact: christian@sanctum.ai

Partner Communities

• Young Federation of German Scientists (Young VDW)

• AI Tinkerer NYC

💡 Editor’s Note: There has been lots of hypes going into this year. AI Agents, world foundation models (yes, the wall-E on GTC), deep research, no-code platform like v0.dev and lovable that make apps for you in minutes. Piling on this, the new image models from Gemini & ChatGPT adds an additional layer of surrealism. All these hype aside, what we can really, reliably do today with AI? We will dive into this topic in our Builder's Corner and show case some actual uses cases that have been quite useful - for me at least : D

P.S. You can find table of contents (TOC) on the left.

🎉 Upcoming AI Events - April 2025

🌟 Featured Events

• [IRL] Manus AI Fireside Chat | Mar 31 | RSVP Exclusive opportunity to meet Tao Cheung, Manus co-founder.

• [Application] Mini-Hackathon v1 | Apr 5 | RSVP Put AI productivity to the test! Go from idea to MVP in just one day with AI Tinkerers NYC.

• [Deep Dive] Cornell Tech Media Tech Summit 2025 | Apr 25 | RSVP Inaugural summit bringing together creators, technologists, researchers, and entrepreneurs to explore AI and XR's impact on media. Four powerful panels cover AI content creation, immersive technologies, trust in digital media, and entrepreneurship opportunities.

📅 Key Events This Month

IRL

• NYC GenAI Companies Showcase | Apr 9 | Apply Meet founders building breakthrough AI innovations, hosted by The AI Furnace.

• Agents are Taking Over NYC with Kamiwaza | Apr 16 | RSVP Evening of demos and networking focused on enterprise-grade AI agents and open-source agent infrastructure.

Application

• Climate Adaptation in the AI Era | Apr 1 | RSVP Brown University hosts Roberta De Vito discussing AI/ML applications for climate resilience and addressing technology's environmental impact.

• Unpacking Ethics in AI with Margaret Mitchell | Apr 2 | RSVP Practical insights on AI ethics through real-world examples from recent generative AI developments.

• Scaling AI with Next Generation Energy Solutions | Apr 3 | RSVP NYU explores how AI growth is reshaping energy consumption and how industries can innovate to meet scaling demands.

• AI Ignition Engine: Business Value Assessment | Apr 7 | RSVP Strategic reality check on where AI delivers value and how to avoid costly implementation mistakes.

• Learn How to Build A Custom Chatbot | Apr 8 | RSVP Hands-on session exploring chatbot applications for exam prep, career development, and business use cases.

Deep Dive

• The US-China AI Contest | Apr 2 | RSVP Dr. Michael Mazarr (RAND) discusses political and technical dimensions of US-China AI competition.

• AI in Production: Video Generation with Modal Labs | Apr 2 | RSVP Hands-on workshop on creating and deploying custom generative video models using open-source foundation models.

• AI: Advancing Foundational Biology | Apr 8 | RSVP MIT/Whitehead Institute symposium exploring AI/ML transformations in biological research with experts from leading institutions.

• Using AI in Qualitative Research | Apr 11 | RSVP NVivo AI demonstration and expert panel on promises and pitfalls of AI in qualitative research methods.

Builder's Corner - AI for productivity... How?

TL;DR. This article showcases productivity examples with reasoning models (basically, models that think) and where it falls short.

Now, I must confess I am a pretty heavy AI user. So far this month I have sent 5.5 Million tokens to o1pro, and used 665 premium requests. So my why view towards whether AI is useful or not could be quite biased : D

So far this year, I have been using AI for creating toys and arts for my toddler, building and launching applications (web & mobile), gathering information and managing those knowledge, organizing my activities and emails, practicing my public speaking skills, refining ideas, managing the NGO and writing this newsletter even.

Did AI save me time and money on this efforts? Yes. But it doesn't always meet my bar for quality...

1. First off, which model?

There are many different types of model today for different modalities: text, audio, image, video. Today, we focus on the text models. Specifically reasoning and non-reasoning models. What are reasoning models?

Now this topic can be expanded to a much longer article on reasoning in models and humans... which I won't digger further today. However, I would refer curious readers to an article on human reasoning and a collection of articles on LLM reasoning.

Now, simply put, reasoning models they are model that 'thinks'. Like o-series models, gemini thinking (2.5), claude 3.7 extended and deepseek r1. So when should I use reasoning models?

OpenAI has a lengthy article about which model to choose and when:

• Speed and cost → GPT models are faster and tend to cost less

• Executing well defined tasks → GPT models handle explicitly defined tasks well

• Accuracy and reliability → o-series models are reliable decision makers

• Complex problem-solving → o-series models work through ambiguity and complexity

Now... if you have no concern on cost, I would definitely recommend using reasoning models for all complex tasks, and 4.5 for tasks focused on contents. Personally I have been using primarily o-series models since it's release and other reasoning models. This means:

• Try o3-mini-high and o1 (or o1 pro if you have it)

• Try Claude 3.7 Extended

• Try Gemini 2.5 & 2.0 Thinking

Last but not least: Don't use DeepSeek R1. According the most recent hallucination benchmark, deepseek-r1 has a whopping ~14% hallucination rate. In contrast, the models above have average hallucination between 1%-2%.

2. Productivity Gains

My productivity gains from AI comes in two forms: AI augmented workflows and AI created workflows. The former refers to when AI is directly tied to the core logic of my task, while the latter is when AI is used to create tools that handle the core logic of my task - this are what AI alone cannot do.

Now, I will examine the efficiency of AI through the creation of the newsletter. I hope it won’t be too surprising when I mention that this newsletter was created with the help of AI. Can you guess which part? : D

There are literally thousands of AI events happening every single month across the nation - the big question is which ones are worth going? Now, this varies from people to people, but there are at least two common criteria: 1. Relevant 2. High Quality.

The initial step shown in Figure 1. was usually done by hand - selecting the events that meets the criteria. For this month, I have tried two Agentic AI tools to help, both failed to different degrees.

• OpenAI Operator: Found 1 event and stopped. (Example)

• Manus: Found 22 events, 3 were usable. (Example)

So for now, step 1 will continue to be manual. However, step 2-4 could be reliably delegated to AI!

I have done all of these steps manually, before reasoning models were available, so I can tell, definitively, how much time AI has saved me : D

2.1 AI Augmented Workflows (Time Saved: 3 Hr/Month)

For the flow, initially, the event information are collected into a google sheet, this creates a large csv file with tons of information. Previously I had to clean up this information manually (as you can see, non-reasoning models don’t quite work). However, with reasoning model, I was able to get much reliable results.

Specifically, AI augmented workflows helped me through step 2-3 and 4B.

The top left of figure 2 is the raw event information collected manually. The top right, is the formatted CSV files using ChatGPT o1-pro. The bottom left is the output copied back to the google sheet. Prompt is simple and direct as below.

Note: For the intricacies of prompt engineering, it’s components I will refer readers to Chapter 5 of AI Engineering by Chip.

Prompt:

<system> output in CSV format </system>

<task>Reformat the events information I have from @Events into the csv header format: Title Cost StartTime EndTime EventType Location Link Description Affiliation</task>

<Events>…..</Events> <Example>…</Example>

Now, I wouldn’t be a geek if I don’t compare this methodically. So here we are:

• Non Reasoning: 4O Result, Gemini 2.0, 3.7 Sonnet, Deepseek V3

• Reasoning: O1 Pro, O3 Mini H, Gemini 2.0 (T), 3.7 Sonnet (T), Deepseek R1

As you can see in the results, non reasoning model all messed up the start/end time, which should be in the following format “04/08/2025 14:00:00”. Reasoning models generally handled this better (O3 Mini H thought really hard but failed.. could be a bug though) .

Next up, the events section above was created using Claude 3.7 as part of step 4B. With the following prompt in one go

Prompt:

Help me draft an event newsletter (in markdown) based on @events . I need very concise, to the point event description, with links to RSVP for the events. Each event should take no more than 2 lines. Example row: Name, Date, Link, Target Audience, Content

<events>…</events>

With reasoning model, it is much less of a concern that the model will omit key information - and we can be very demanding with our tasks.

2.2 AI Created Workflows (Timed Saved: 1.5 Hr/Month)

Now, there are tasks that AI cannot do. Specifically, in the context of this newsletter, it’s creating the events on Google calendar (Bottom right of Figure 2).

The entire flow is quite simple actually. Without AI, I would have set up automation tools like Zapier or Make to create a workflow like below. The downside? I need to PAY for it!

Now, of course, I could pay for it, but for such a simple task like this one. I have absolutely 0 interest paying ~$10-$20 a month for it. So, I opted to do it myself and get the function for free.

The workflow is quite simple, as show in Figure 4. You can first write a prompt like the one below to generate the initial AppScript code, and then iterate on it.

Prompt:

I have a google sheet with the following columns [Title Cost StartTime EndTime EventType Location Link Description Affiliation]. I want to create a google calendar event for calendar “$myemail@gmail.com”, the event title should be based on Title, the link should be Link, the location should be Location. Now, the description should be a summary of the [Cost EventType Description Affiliation], and I want the summary to be written using OpenAI api. Prompt user for api key.

<task>Write a AppScript to finish the task </task>

Check out the initial code from the above prompt. This usually takes just 2-3 iterations to get it down. Since reasoning models are much much better at leveraging contexts, it’s best you throw in large contexts and then reference them in the prompts - the model will go and figure it out.

In short, using AI to create this workflow have saved me 1.5 hours/month if I were doing this manually, or ~$10-20/month if I pay for automation tools. Doesn’t sound like a lot right? But this is just one simple use case and you could create more!

Now, this is NOT how you should be coding for even moderately complex projects, where the issue becomes context management, which is a common issue when coding using cursor and other tools (Repomix is a good tool to bundle all context in one go btw).

Remember: With AI, you no longer need to pay a fortune to perform all of the simple tasks - email summary, news summary, event creation, calendar management etc. Just DIY - for free. If you are interested, I can cover these examples in future posts : D

3. Parting Words

In this post we have walked through how to efficiently use reasoning model through a few examples. This is only the beginning of what we can do with AI today.

One particular issue I mentioned in section 2, is that

AI tools are inherently probabilistic and quality is a concern

This still holds true for all of the tasks I have described above. I still need to manually verify the event links created, the time, content description for all the events created. I need to verify if the code generated is correct, and iterate through testing.

To learn more, one person I would definitely follow is Andrej Karpathy (ex-OpenAI, now Eureka Labs). This video show quite a few use cases that could hopefully spark some ideas.

Stay tuned—next time, I’ll reveal the specific tech stack that powers these projects, including how I integrate tools like Cursor, v0.dev, and beyond!

Next up in community

We just wrapped out Educational event with Newtown Public High school (Thanks to Google). In Q2, we will host several more events, so stay tuned! !

💡 Editor's Note: This month, we're diving into the intersection of AI with healthcare, materials science, and human-centered design. With exciting talks from institutions like MIT and Cornell, there's something for everyone in the AI community. Our focus continues to be on practical applications that create real impact. The highlight of February was our RAG workshop with Steven Bonilla from Microsoft—details below for those who missed it. Remember, our subscribers get first access to all upcoming event invitations!

🌟 Featured Events

• Human-centered AI Beyond Human-in-the-Loop with Prof. Dr. Przegalińska (Mar 20)

• AI for Social Good Innovative Applications at NYU with Prof. Dr. Ruopeng (Mar 4)

📱 Stay Connected

• Subscribe to event updates

• Sign up here if you would like to speak at our events (or any other communities).

📅 Key Events This Month

Featured Event: Human-centered AI

• Human-centered AI: Why a Human in the Loop Is Not Enough

  • Date: March 20, 12:00 PM

  • Speaker: Aleksandra Przegalińska

  • Focus: Moving beyond standard frameworks toward transformative human-AI partnerships

  • Register Here

Technical Focus: AI Community NYC Meetup

• AI Tinkerers NYC Meetup - March 2025

  • Date: March 4, 6:00 PM - 9:00 PM

  • Location: NYC (In-person, details upon registration)

  • Features: LLM demos, sponsor showcases (SignalFire, Comet)

  • Register Here

• NYC AI Users - AI Talks, Demo & Social

  • Date: March 25, 7:30 PM - 10:30 PM

  • Location: NYC (In-person, details upon registration)

  • Features: Fine Tuning Gen AI and ML & LLMs

  • Register Here

Special Event: AI for Social Good

• AI for Social Good

  • Date: March 4, 4:30 PM - 6:00 PM

  • Speaker: Dr. Ruopeng An (NYU)

  • Topic: Projects addressing COVID-19 disinformation, tracking public sentiment, and more

  • Register Here

📆 Additional Events

Week of March 3-9

• NYC GenAI Companies Demo Day (Mar 11, 6:00 PM)

Week of March 10-16

• Gesture-based Human-Machine Interaction (Mar 13, 12:00 PM)

• AI in Healthcare: Opportunities and Risks (Mar 13, 12:00 PM)

• MIT.nano Seminar: AI for 2D Materials (Mar 17, 3:00 PM)

Week of March 17-23

• Human-AI Ecosystem for Health and Wellbeing (Mar 21, 11:00 AM)

Week of March 24-31

• Fine Tuning Gen AI and ML & LLMs (March 25, 07:30 PM)

• Privacy Challenges in the Age of AI Agents (Mar 27, 12:00 PM)

• Climate Adaptation in the AI Era (Apr 1, 3:00 PM)

[Nexus Recap] RAG: Building Real-World AI Applications

Now, if you would like to learn more about the event you can find the recording, slides, and relevant resources below!

Last Friday, we had the privilege of hosting Steven Bonilla, an LLM Engineer from Microsoft, who gave us an amazing presentation on industrial grade Retrieval-Augmented Generation (RAG) in chatbot applications for financial analysts!

Steven walked us through the basic concepts and architecture of a RAG system (our last post also briefly touched on this topic) , as well as key components used in the practical designs such as:

• Contextual Query Processing: Creating awareness of conversation history

• Named Entity Recognition (NER): Identifying key information for retrieval

• Retriever Router: Deciding which data sources to query

• Scaling Techniques: Batching calls and using appropriate model complexity

One challenge presented was similarity calculation between query and contexts - you would not believe how many of these implementations just uses simple vector retrieval. Alas, there are more solutions on the horizon for better and more accurate ways to find relevant contexts (i.e Gutiérrez, et al., 2025).

In short, these are the real world considerations when leveraging RAG in your products - we cannot always operate on infinite budget (seriously, a big concern) and have to use less powerful models to deal with our problems. Yet,

"There's always a way - if you're committed." - Tony Robbins.

Technical Deep Dive

For those interested in exploring RAG further, check out these resources:

• Technical Guide to Vector Databases by Google

• Evaluation Frameworks for RAG Systems by Openai

• Cost Optimization Strategies for LLM Applications by Amazon

I have put more resources in the shared drive, feel free to check it out (simply subscribe)!

Again, we want to extend our heartfelt thanks to Steven for sharing his expertise and to all who attended!!

Subscribe now

Curated with 💙 for the NYC AI community

💡 Editor’s Note: I have been comparing various models in the past months and extensively on Gemini 2.0 and O1-pro since it came out. The performance on long contexts for these reasoning models are amazing - and how to structure instructions is also quite interesting. That being said, there are still limits on the context and finding the right information is still the key to boost performance for both non-reasoning models and reasoning models. I have compiled a list of readings towards the end of the article for those who are curious!

Join us to learn more about RAG on Feb.28!

On February 28, Steven Bonilla, an LLM Engineer at Microsoft, will share how he built a RAG-powered chatbot for financial analysts. From architecture to best practices, his insights will help you leverage RAG to harness AI for real-time, high-impact applications.

Sign up today at https://lu.ma/3e0y47ei !

The era of huge contexts - How to give bot what it needs?

AI models, especially large language models (LLMs), can technically take in a lot of context – Gemini boasts context windows long enough to fit whole books or even an encyclopedia​. But just because we can give an AI all our information at once doesn’t mean we should.

Why?

I still remember few years ago when I tried to explain my research to my lab mates - a rather complicated effort with lots of maths, programs and psychology concepts intertwined. Back then I was still rather in experienced in presentations (still improving today, of course) and I showed every single detail of the project. At the time I felt pretty good, but of course, much of my friends didn't quite follow my train of thought and was overwhelmed by the large amount of information I threw at them.

Sound familiar? Turns out giving too much information all at once is not really a good way for presentation - or just about any conversation. Just like my lab mates and Profs who got lost in my sea of details, AI models can get overwhelmed, too!

Aside from the large amount of contexts, there comes time where we cannot upload information to public AI chatbots too - privileged research data, unpublished manuscript, client-specific information etc. but we still want to leverage the convenience of LLM. (On that note, make sure to turn off data sharing when using public chatbots!).

In short, too much data, private data source, regulations etc, all makes it difficult to us to provide AI all it needs to perform its task - even today.

Deep Dive: Interestingly, the newer reasoning models does quite well (Databricks, 2024). However, such is not the case for all models. Older models like GPT-4o (yes, older. this is how fast things evolve : D) was not quite good at handling long contexts, utilizing ~10-20% of the tokens supplied (Kuratov et al., 2024). OpenAI’s o1(pro) and o3 models accepts up to 200k tokens, while Gemini boasts 2 million tokens (2.0 thinking), in contrast deepseek-R1 has a limit of 128k.

So... what can we do?

One of the methods that we could use is Retrieval-Augmented Generation (RAG). In real life, we humans don’t memorize entire textbooks (well, unless we have an exam coming up tomorrow, then we cram it all! And, of course, forget everything soon after), we just look up what we need at the moment, that is most closely related to our problem.

RAG does the same thing for AI. It pulls in only the relevant bits of information (chunks of contexts, think how you chop up a baguette - it would be rare to see someone eat it whole) right when the bot needs them. This way, the AI isn’t juggling a mountain of irrelevant details; it’s laser-focused on the stuff that matters. The result? Fewer mistakes, clearer answers, and a much happier bot (if bots had feelings, of course!). Now, there are many complex terms in Figure.1, which I recommend check out Chapter 6 in AI Engineering (Huyen, C., 2025).

Now, getting relevant chunks optimally is an interesting challenge in itself - Anthropic has a good article on contextual retrieval (Anthropic, 2024). Think of it as 'decorations' to help you find items easier. Like the name stickers for my kids diapers at day care - parents often buy the same brands, and it would be quite difficult to tell it apart without some form of name tags on it!

Then when should we consider it?

Figure.3 provides one possible decision process for choosing RAG. Now, the reality can be much nuanced because there are many practical considerations such as modality (if you need multi-modal model), cost ($$), regulation, technical expertise of your team (can you maintain a local RAG set up in the long term).

If you do choose to use RAG, and would like to learn more about the trade-offs and design considerations, then join our study group for a in-depth discussion on the topic from Steven Bonilla@Microsoft!

Up Next

Stay tuned for our next “Builder’s Corner” article: Working with Thinking Chatbots: How to Speak Simply and Directly, where we’ll explore how you can leverage reasoning models effectively and with lots of contexts. Seriously, how can I tell the model where to look? Btw, be sure to check out ‘Reasoning best practices’ by OpenAI (2025). The article says just keep your prompt ‘Simple and Direct’ - a feat not as so simple as it sounds.

What’s the last time you were frustrated because you couldn’t get your words across to one another? If we can all just speak ‘simple and direct’ life will be much easier - and there won’t as much misunderstandings / miscommunications.

It is the same with reasoning models. Oftentimes I would find myself describing a task to the model, then instantly realize what I have said could be misinterpreted - and not surprisingly it often ends up misinterpreted... If you have similar struggles with reasoning model, please comment below!

References & Reading List

1. Databricks. (2024). The long context RAG capabilities of OpenAI o1 and Google Gemini. https://www.databricks.com/blog/long-context-rag-capabilities-openai-o1-and-google-gemini

2. Kuratov, Y., Bulatov, A., Anokhin, P., Rodkin, I., Sorokin, D. I., Sorokin, A., & Burtsev, M. (2024). Babilong: Testing the limits of LLMs with long context reasoning-in-a-haystack. In The Thirty-eight Conference on Neural Information Processing Systems Datasets and Benchmarks Track.

3. Anthropic. (2024). Introducing contextual retrieval. https://www.anthropic.com/news/contextual-retrieval

4. Huyen, C. (2025). AI engineering: Building applications with foundation models. O'Reilly.

5. OpenAI. (2025). Reasoning best practices. https://platform.openai.com/docs/guides/reasoning-best-practices

💡 Editor’s Note: In the coming weeks, we’ll be delving into critical conversations on AI safety with leading figures such as Geoffrey Hinton! Beyond the buzz of impressive generated videos and images, our focus will be on harnessing AI’s power to address real-world social challenges. That’s why we’re launching our Social Good Hackathon and Incubator next quarter, designed to spark innovative projects that make a tangible difference in people’s lives. Stay tuned, and remember that our subscribers will be the first to receive exclusive event invitations!

🌟 Featured Events

• Autonomous Armageddon Nuclear Weapons & AI with Geoffrey Hinton (Jan 26)

• OpenAI X AI Tinkerers Hackathon Realtime Voice x Reasoning (Feb 15 & 16)

• The Peacemaking Machine AI & Democratic Deliberation (Feb 27)

🌟 Google’s Social Good GenAI Accelerator!

• To quote Google Director David McLaughlin

We're looking for the next generation of AI innovators to join our Google.org Accelerator: Gen AI program. Selected organizations will participate in a six-month accelerator that includes technical training, Google Cloud credits and deep mentorship from the best of Google and our broader mentor ecosystem, along with a share of $30 million in funding.

Check out our new video for an overview of our first class and to hear from current participants: https://lnkd.in/gR7nU9qk

Don't miss out on this opportunity to solve world-changing problems using Gen AI. Apply today!

Coming up Next: Builder’s Corner

Hi friends! Nick here.

Coming into 2025, in addition to the events, I will be sharing insights on AI usage from my experience as an AI developer and neuroscientist. These posts will have a different tag, so you can skip them if you want to : D

In the next session I will cover my experience and observations tuning the popular tools (ChatGPT o1/o1pro, Claude, Cursor) with prompt engineering, amazing libs from our friend which turns cursor to a multi-agent platform.

If there are any tools you would like to see, comment below!

📅 Key Events This Month

Featured Event: Nuclear Weapons & AI

• Autonomous Armageddon: Nuclear Weapons and AI

  • Date: January 26, 2:00 PM GMT

  • Speakers: Geoffrey Hinton, Connor Leahy, Dr. Ruth Mitchell

  • Focus: AI integration risks in nuclear systems

  • Join Webinar

Technical Focus: OpenAI Hackathon

• Realtime Voice x Reasoning Hackathon

  • Dates: February 15-16

  • Time: 8:00 AM (Sat) - 3:00 PM (Sun)

  • Location: NYC (In-Person)

  • Sponsor: OpenAI

  • Register Here

Special Event: AI & Democracy

• The Peacemaking Machine

  • Date: February 27, 4:00 PM EST

  • Speakers: Michael Henry Tessler & Michiel A. Bakker (MIT/Deepmind)

  • Topic: AI-enabled democratic deliberation

  • Join Discussion

📆 Additional Events

Week of February 4-9

• NYC GenAI Demo Day (Feb 4, 6:00 PM)

• Multimodal Model Workshop (Feb 6, 4:00 PM)

Week of February 10-16

• Quantum Processing in the Brain (Feb 12, 1:00 PM)

• Embedded Ethics Seminar (Feb 12, 1:00 PM)

• AI in Waste Management (Feb 12, 2:00 PM)

Week of February 24-29

• AI & Freedom of Thought (Feb 25, 12:30 PM)

📱 Stay Connected

• Sign up at our website if you

  • would like to speak at our events or any other tech communities!

  • would like to sponsor one of our events

• Subscribe to event updates

Subscribe now

Curated with 💙 for the NYC AI community

💡 Quick Take: With the new AI products releases coming in at the end of the year, I’m sure many of you are as excited and confused as I do - what can I do now? where are we headed? It’s hard to say for sure, but agentic AI is definite up and coming!

💫 2024 Recap & 2025 Outlook

As we bid farewell to 2024, we witnessed remarkable advancements in AI development - from multimodal breakthroughs to enhanced reasoning capabilities, to agentic AI and amazing text-to-video models. But 2025 marks one pivotal shift: The Rise of Agentic AI.

What previously took months of development can now be accomplished in weeks with AI assistance, and potentially days with Agentic AI. However, it's crucial to understand that current Agentic AI typically gets us 60-70% of the way to a minimum viable product. The remaining 30-40% - the refinement, optimization, and deep integration - still requires human expertise and experience.

This is why 2025 is the Year of the Builder. Success lies not in relying solely on AI, but in skillfully combining human insight with AI capabilities. Our mission is to empower you with the knowledge and tools to thrive in this new paradigm!

🌟 This Month’s Highlights

[Special Partnership] Learn about what is AI and how it works with Superlinear Academy (10% off for subscribers).

[Featured Talk] AI and the Future of Lawmaking: Innovation at the Library of Congress with Natalie Buda Smith (Jan.23rd)

📅 January Calendar

Application Series: Leveraging AI to Solve Your Issues!

1. [Jan 9, 19:00 EST, MIT] AI and Trust | Virtual

2. [Jan 10, 12:00 EST, MIT] AI revolution in HR: Unveiling skills in the modern job market | Virtual

3. [Jan 16, 12:00 EST, MIT] Mitigating the danger of AI | Virtual

4. [Jan 21, 14:00 EST, MIT] Aligning Innovation and Equity in the Digital Economy | Virtual

5. [Jan 22, 12:00 EST, MIT] Data deletion in AI systems: The illusion of erasure and its privacy implications | Virtual

6. [Jan 23, 17:00 EST, Northeastern] AI and the Future of Lawmaking: Innovation at the Library of Congress with Natalie Buda Smith | Virtual

Beyond AI: Get Interdisciplinary insights!

1. [Jan 30, 18:30 EST, Columbia University] A Fireside Chat with Anthony Fauci, M.D. | Virtual

🤝 Community Spotlight

• Thriving Community: Join 1,000+ researchers, professionals and builders!

• Growing Contents: Past sessions and promotions.

  • Subscribe and you will gain access within 24 hours

• Tell us events you like!

What do you want for 2025?

🛠️ Builder's Corner: Spinning Up Your Local AI Server

From our previous workshop with Sanctum.ai and Dr. Yan Wang from Superlinear Academy, many of the audience asked about how to set up a local AI server.

This may sound like a daunting task - having to navigate through the models, the stacks, the web UI, the API …… But worry not! What would have taken days for you now takes minutes! I have taken the liberty to create a walk through on setting up local AI servers.

So… what exactly did I do?

• I instructed cursor (an AI-powered IDE) to set up a local AI server for me using ollama and openwebui, recording it’s actions in development record

• It set up both tools and tried to download mistra (my mac is not sufficient to support), so I told it to use llama - which turns out it doesn’t work too

• I relayed the insight to cursor, and it helped me choosing tinyllama

• …and it worked!

The entire process took me ~12 mins, 5 of which were spent on downloads. Video editing actually too me ~40 mins, and music was generated using suno! To save your time, here are my prompts:

> Create development record. I want to spin up a local webpage for AI models. I want to use openwebui + ollama, provide a full implementation plan in the record
> proceed (X2)
> start with llama
> start serving the model

Occasionally AI would slack off, in which case you would need to tell it to go on : D Now, there are more tricks that I personally use, which might warrant an article of its own. One tip I would give though, is always use github (or any version control) when using agentic AI - so when it messes things up, we can roll it back confidently!

📱 Stay Connected

Subscribe now