In November and December of last year, my co-founder Tomas and I didn't exchange a single message.
That might not sound like much. But we're best friends. We've been building together for years. The kind of relationship where you're texting at midnight about some idea you can't let go of, or sending a voice memo while driving because you just figured something out.
Two months of nothing. That's how I knew things had gotten bad.
What happened before
About eighteen months before that silence, we were close to landing a contract that would have changed the entire trajectory of what we were building. We were in the blockchain space, building real enterprise infrastructure, not hype products. We had a Fortune 500 company on the table.
Then their leadership changed. New CTO. New priorities. The plan got nuked. That's it. That's how fast it happens.
What followed was eighteen months of trying to figure out what comes next. We stayed in the original space for a few weeks because that's what you do. You convince yourself there's still something there. There wasn't. The market had disappeared. We needed to move on.
Our technical co-founder was gone. Our third partner eventually left because we were spinning our wheels. I was the only one left who could build anything, and I'm not a full-stack engineer. I'm the guy who hacks together a proof of concept to figure out if something is possible. I find the gap. I don't ship the polished product.
So we had runway from an existing contract we had to fulfill and no clear path forward. We tried ideas. A lot of ideas. All of them went nowhere.
During those two months of silence, everything that had been building up finally surfaced. The stalled momentum, the dead ends, the identity I'd built around always moving forward. I know now that was months of depression catching up to me all at once. Not the metaphorical kind. The kind where you stop talking to your best friend for two months because you have nothing to say and everything feels stuck. My identity has always been tied to momentum: building, forward motion, knowing where I'm going even if the path is messy.
What changed
Sometime around early 2025, the AI tools got good enough that I could actually build things. Not just hack together a demo to prove a concept. Actually build them. The gap between "I can see that this is possible" and "I can put something in someone's hands" closed in a way that it never had before. For the first time, a small team without a traditional engineering lead could build and ship real products.
I picked up some contracting work and for the first time in a while, I was shipping things I was proud of. Building with confidence instead of duct-taping things together and hoping nobody looked too closely.
And then Tomas resurfaced. Or maybe I did. Honestly I don't remember who texted first, which is kind of funny for two people who used to talk every day. But he could tell something had shifted. I was working again. Not spinning. Not brainstorming into the void. Actually making things.
Here's the thing about Tomas: he's one of the sharpest product minds I've ever been around. He's been deep in Reforge-level thinking since before I knew him. Aha moments, core loops, retention mechanics, engagement cadences. The stuff that separates a product someone opens once from a product that becomes part of how they work. He sees those systems in everything. But all of that knowledge, all of those frameworks, they're academic if you can't actually build the thing underneath them. For a long time, we couldn't. He knew exactly what a great product should look like and we had no way to make one without a full engineering team.
Now we could. And something clicked for both of us at the same time. He has something he's always told me: "I don't work well alone. I need to be thinking things through with someone." For two months, neither of us had that. We were both sitting in the same silence from different rooms. Once we started talking again, it was like the whole machine turned back on. I could build, he could shape what to build, and the tools had caught up to where we needed them to be.
We weren't just two people with ideas anymore. We were a product strategist and a builder, both sharper than we'd been eighteen months ago, with tools that didn't exist back then.
The old version of us would have picked one idea, raised money, hired a team, and spent two years trying to make it work. That's how startups are supposed to go. But that model assumes you need fifteen people to build and operate one product.
The model
What we landed on is closer to a venture studio than a traditional startup, but the mechanics are different from what that phrase usually means.
A team of three people building and operating a portfolio of vertical-specific products on shared infrastructure, using a repeatable system that gets faster every cycle.
Every market that looks like one big addressable opportunity is actually dozens of smaller ones underneath, split by how people actually work. A private chef running weekly meal prep for four families has almost nothing in common with one doing pop-up dinners for events. A small real estate investor managing contractors across five rehabs doesn't need the same tool as an enterprise brokerage. These segments were never worth building for before because the cost to build, operate, and support a product for each one required an entire company.
AI didn't just make engineering faster. It compressed the entire operational loop: building, testing, iterating, supporting, documenting, communicating with users. Three people with the right infrastructure can now do what used to require twenty.
But not just for one product. For many, in parallel, each targeting a different wedge where the pain is sharp and the users are reachable.
Each venture follows the same loop: find a problem through direct observation, enter through the narrowest viable wedge, build the smallest useful product, get it into real hands fast, iterate against real feedback. What compounds isn't just the products. It's the operating knowledge. What insight led to a real opportunity. What wedge converted. What messaging worked. What users actually valued versus what they said they wanted. The tenth venture runs at a fundamentally different speed than the first.
We're not picking verticals from a spreadsheet. We're following problems. The wedge is determined by where the pain is sharpest, distribution is most accessible, and willingness to pay is real.
What stays human
This is the part that matters most to me.
AI handles building, operating, and iterating. What becomes more valuable, not less, is being in the real world. Noticing what breaks in someone's day. Earning trust. Interpreting the feedback that doesn't fit neatly into a form field. Making judgment calls about what to build and what to ignore.
The best product insights come from observations that don't exist on the internet. A conversation with an operator about why every tool they've tried fails. Watching how someone actually runs their workflow versus how they describe it. Those data points are scarce, can't be scraped, and are the raw material the whole system runs on.
The goal is a small team that spends most of its time collecting signal in the world, while the machine behind them builds, ships, operates, and learns.
That's not a pitch. It's how we're actually working right now. We have products in the pipeline: a private chef assistant, an events product, others in various stages. The system exists. It's early, and we're still figuring out a lot of it in real time.
Why I'm writing this
Most mornings I wake up and somewhere between the coffee and the laptop I think: fuck, is this actually going to work?
There's a Mac Miller tweet I keep a screenshot of framed on my desk.
Stop keeping score. Just keep swimming.
I spent eighteen months keeping score, measuring where I was against where I thought I should be, against the deal that fell through, against the version of success I was attached to. That scorekeeping didn't produce anything. It just made the stuck feeling worse.
What actually moved things forward was the same thing that's moved everything forward in my life: curiosity, pressure, uncertainty, and then a fuck it decision to build anyway.
I'm not writing this because I have it figured out. I'm writing it because I think the way software gets built and businesses get started is changing in a fundamental way, and I want to be part of that conversation. I also think more people are going through some version of what I went through — the identity crisis that comes when the thing you built stops working and you have to figure out what's next — and it helps to hear that it's messy and it sucks and you can come out the other side.