I’m delighted to share my chat with Gary Chou, a long-time player in the NYC tech scene. He’s maybe best known as the founder of Orbital, a cozy coworking space / incubator that’s now an online community of practice built around quarterly reflection (I’m a happy member). He also spent time facilitating community at Union Square Ventures. He co-taught the legendary Entrepreneurial Design class at SVA, where students had to not only design a product but successfully raise $1,000 for it on Kickstarter. And he’s currently working on a new non-profit.
Gary joined me to share his experience using AI as a collaborator, running communities of practice for product leaders, and helping students find their product sense.
Press the big button above to listen here / in the Substack app, or watch the video on YouTube. Highlights below!
The benefits of role collapse
Tangible concepts. Just like Gary’s $1K challenge that brought product and market processes together for greater insight, AI’s instant prototypes allow the complexities of tech to be seen, felt, and refined much sooner.
Better decision making. Expanded capacities and accelerated timelines mean we can build, measure, and learn with much more sophisticated understanding.
Flexible, creative teams. No more silos. Everyone could show up to a meeting with prototypes reflecting their unique expertise, already debugged by virtual testers.
“The thing that I’m most excited about is just: as an individual, if you have an idea, you have access to this tool that lets you get way farther with a lot less capital than you would otherwise need. And that creates a lot of potential for new and interesting ideas to emerge.”
—Gary Chou, Orbital
A modern process for solo builders
Work in steps, not one shot. AI prototyping tools like Lovable and V0 work best when tasked with one specific problem at a time, like the mobile navigation instead of the whole app.
Role-play team members. Ask the AI to play designer, PM, or engineer — hopefully using your process knowledge from working on classic tech teams to know how and when to consult them.
Build virtual SMEs. Things like tax code can be fed into LLMs to create subject matter experts that are good enough for rounds of testing. LLMs can do code review on other LLMs to overcome the shortcomings in each.
Achieve quality through quantity. AI-assisted drafts will be lower quality, but higher quantities of iteration can more than compensate.
Keep humans in the loop. We can’t outsource good judgement. Students (and all of us) still need a variety of social skills.
“ We’re gonna have to rewire how we engage with other people. Whether they’re our teammates, clients, or collaborators… [AI] opens up a lot more creativity in how humans can organize with each other beyond the industrial models that we currently have.”
—Gary Chou, Orbital
Avoiding the pitfalls
Have a plan. AI is “too crude and powerful” of a tool; you need a framework (e.g. team collaboration models) to recognize and resolve issues.
React, don’t rely. Thinking with LLMs is more of a “pinball model”, bouncing ideas back and forth. Its unreliable outputs are like messy prototypes, helping you slowly get closer to what you want to say.
Test the magic. As we become instantly prolific creators, infatuated with the power of our new tools, we can lose our sense of judgement. Outside user testing is more important than ever.
Rest and reflect. The pace AIs enable is also a trap. Know when to step away, decompress, and reflect.
“There’s this period of getting way too drunk on the quick iterations. You feel like, oh my gosh, there’s this magic happening and this is the best thing in the world. And then you go test it and people are like, I have no idea what you’re talking about here.”
—Gary Chou, Orbital
Thriving in the new age
Everyone is adjusting. Students, you’re not alone. Even senior people are struggling to understand what to do next.
Try something small. Learn through doing. Work in small steps.
Find a community of practice. It’s easy to get “lost in the chaos” when you’re solo; find a forcing function for reflection and vulnerability and growth.
“I think there’s gonna be a lot more business opportunities for smaller teams to pursue things that were previously labor constrained... There’s a lot of things that are now financially viable, and I think within that, there’s a whole lot of learning opportunity that people can get.”
—Gary Chou, Orbital
Links
Tools
Lovable — AI-powered full-stack app builder
V0 by Vercel — AI UI generation tool
Airtable — no-code database used for workflow automation
Claude (by Anthropic) — used for prototyping, coding, and reasoning through complex design questions
Books
(My rec:) Surely You’re Joking, Mr. Feynman! by Richard Feynman: an inspiring autobiography of nonstop creative thinking
Keep up with Gary
His blog: garychou.com
Lessons learned from teaching the $1K Challenge (with Christina Xu):
Orbital membership (online community of practice): orbital.nyc/member
