About two weeks ago, my friend Varun threw a farewell party. He’s moving to the Bay Area. The apartment was full of researchers, founders, and the kinds of people who get it when it comes to how much AI is about to turn everything on its head, the kind of people where every conversation makes you think I want to talk to this person again.
I left the party with a familiar feeling. I’d just spent an evening with exactly the kind of people I want more of in my life, and I knew that, in all likelihood, I would never see most of them again.
We exchanged a few numbers, sure, but I’ve done this enough times that I know follow ups are rare and when contemplating the cognitive overhead of coordinating two people’s scheduling and location preferences, I usually just file it under the “I’ll think about this later when I have bandwidth” which usually never happens.
This time though, I wanted to try something different. This felt like a software-shaped problem, so, a few days later, I started building.
Epistemic Density
There’s a reason I live in NYC and not the Bay Area. I like having at least some friends with interests outside of technology. But I also sometimes want a particular kind of conversation. I have Claude, ChatGPT, and Gemini subscriptions and actually know what the latest models are capable of and how fast they’re improving. I’m building forecasting bots, experimenting with agents, and working on research. I spend my days thinking about how the world is about to change, and I want to talk about it with people who are thinking about it too.
People talk about the Bay Area as the place to be for this, and there’s something to that, but I think it’s less about geography and more about what I’d call epistemic density: the odds that the person next to you at a coffee shop has updated on what AI means for everything.
In New York, I’m surrounded by brilliant people, but most of them don’t spend their weekends building agents. That’s fine. They have rich lives and I appreciate the conversations I have with them, but I also want conversations where I don’t have to start from scratch explaining why this matters.
My friend Ben Turtel, who runs an AI startup here in NYC, put it perfectly: his friends treat the things he says about current reality as predictions… that are wrong! That’s my experience too. I want to talk to more people who “get it”, but, right now, it feels like they’re scattered around the city and I don’t know how to find them. Maybe they’re not so scattered though. Maybe they’re just around the corner and we just haven’t found each other yet.
Closer Than You Think
Steve Jobs said “If you are willing to travel around the world to meet a teacher, one will appear next door”.
The people I want to know are often hidden in plain sight. Varun lived five minutes away, yet we were strangers until we met at a Dwarkesh Patel party. Another friend lives even closer, we’d known each other for years, but we only started getting coffee in the last few months after discovering we lived in the same neighborhood at a conference on the other side of town. I have multiple other NYC friends who I only really got to know from events in the Bay Area.
Reducing Friction
The concept was like Calendly, but for coffee meetings. I started working on coffeemeet.xyz the next day, made it live after a few days of testing, and by the following weekend, I was already meeting someone who booked time with me.
The concept is simple: I pick my favorite coffee spots, I set when I’m free, and anyone can book a time to meet me.
For my profile, I picked coffee shops nearby. This isn’t because my time is more important than anyone else’s. It’s because between work, teaching, forecasting, and life with my partner I don’t have many free time slots. If meeting someone new requires an hour-long commute, I know from experience that “let’s do this again” turns into “maybe next month” turns into nothing. The friendships that actually develop are the ones where getting together is easy. Varun was five minutes away and we still only managed coffee once or twice a month. Proximity is what makes consistency possible, and consistency turns conversations into friendship.
Software-Shaped Problems
Here's the thing: the gap between an idea and a working tool has collapsed. Friday night, it was just a feeling at a party. A few days later, it was an app people were using.
I could have built coffeemeet without AI, but “could have” and “would have” are different things. AI didn’t spit out a completed app from a single prompt for me, but it compressed a project that might have taken a couple of weeks into a few hours. And that speed changes the calculus on what’s worth building. Problems that used to feel too small or too personal are now worth a Sunday morning.
I think a lot of people are sitting on little problems like this, software-shaped holes that could be solved if they knew where to start. Not every idea needs to be a startup. Some just need a clear idea and a few hours to work out the kinks.
Come Have Coffee With Me
If you’ve read this far, I want to meet you.
That’s the whole reason I built this and the whole reason I’m writing about it.
If you’re in or around Brooklyn, or willing to make the trip, pick a spot and a time on coffeemeet.xyz. I’d love to talk about forecasting, AI, what you’re working on, what you’re reading, or whatever’s on your mind.