January 2026
In the summer of 2025, my cofounder and I parted ways at Epistemy.
We'd built something real together — 28k MRR, clients across six countries, a 14-person team. But after a year of working side by side, we both knew it: we just didn't enjoy working together that much. The company was growing. The partnership wasn't.
This was the second time I'd lost a cofounder. The first was in Singapore on my previous AI venture. That time, I came back to Europe with nothing. This time, at least I had a company still running. But I was back to square one on the thing that matters most — who you build with.
So I applied to Entrepreneur First in London.
EF puts ~80 ambitious people in a room and gives you a few months to find a cofounder, validate an idea, and pitch an Investment Committee for a $125K SAFE. It's structured cofounder dating meets startup boot camp.
I came in with more traction than most — eMatura was doing €500K ARR across 11 markets, Epistemy had real clients and revenue. But none of that matters if you don't have someone to build with. That's what I was there for.
The first few weeks at EF are basically speed dating for founders. You do short working sprints with different people, test chemistry, and move on fast if it's not right.
I explored a few potential cofounders. One was a strong AI researcher — great credentials, good self-awareness, but the dynamic didn't click. Another wanted to build localized consumer apps. We spent about four days together before I realized the fit wasn't there — he dismissed concerns too quickly and optimized for speed over truth-seeking. In a cofounder, that's a dealbreaker.
Then I met my current cofounder.
He's an ML engineer who'd built production systems at a fintech unicorn, processing millions of transactions. He'd also previously built a ridesharing marketplace. His previous EF pairing had fallen apart within hours.
The complementary skills were obvious from day one. He builds. I distribute. He's deep on AI/ML and product. I'm deep on GTM, sales, and ops. We started working together full-time almost immediately.
Honestly, the pairing didn't look like a sure thing early on. But you only know if a cofounder relationship works under pressure. And we were about to get a lot of that.
We went through four distinct ideas in about two months. Each one felt promising until it didn't.
Idea 1: Help companies distribute to new markets. Killed it quickly — automating B2B distribution means building AI SDRs (absurdly crowded market), and every B2B company is so different that scaling this becomes a consulting business, not a tech company. One of our advisors put it well: "As soon as the marginal unit of growth is a human being, it's a consulting business."
Idea 2: Help consumer brands through organic UGC. We pivoted to consumer brands and realized they have more standardized playbooks. Started studying AI apps — discovered they scale through organic UGC accounts. But creator management breaks at scale.
Idea 3: UGC marketplace. First attempt at framing the opportunity as a marketplace connecting brands with creators. The feedback we got: "That just sounds like what marketing agencies do." Fair point. Marketplace model doesn't capture enough value. Thin margins. Competing on creator supply.
Idea 4: Testing infrastructure for performance marketers. This was the one. The non-obvious insight: TikTok's algorithm prioritizes fresh accounts over established influencers. This isn't a temporary arbitrage — it's structural. So instead of competing for expensive creators, we build the infrastructure to recruit, train, and coordinate fresh accounts at scale.
Revenue exploded. €250K ARR in five weeks. 80% close rate. 17% cold outreach response rate.
The difference between idea 3 and idea 4 looks subtle from the outside. But the positioning shift — from marketplace to infrastructure — changed everything. We stopped competing for supply and started creating it.
EF's feedback culture is intense. Advisors will tell you your articulation is horrible, your pitch doesn't make sense, and you should stop talking until you figure it out. Some of the sessions were genuinely painful.
Some of the pushback was useful. Getting pressure-tested on your reasoning forces you to sharpen your thinking. When someone catches you being imprecise about revenue numbers, or calls out that you're trying to tell them what they want to hear — that's valuable. It makes you a more disciplined communicator.
But here's what I think people don't say enough about EF: the program can try to box you in.
Advisors have strong opinions. They push you toward frameworks, positioning, and narratives that fit what they think an investable company looks like. And if you're not careful, you start optimizing for what the advisors want to hear instead of what you actually believe is the right thing to build.
I've seen it happen to people in our cohort — genuinely great founders who started second-guessing themselves because the feedback was so relentless. Some fizzled out not because their ideas were bad, but because they couldn't handle being told they were wrong five times a week. The irony is that the traits that make someone a great founder — conviction, stubbornness, independent thinking — are exactly the traits that EF's pressure cooker can grind down if you're not careful.
What helped us the most wasn't the feedback itself. It was having enough experience from previous companies to know when the feedback was right and when it was just someone's opinion. If this had been my first startup, I'm not sure I would have had the filter to tell the difference.
There was also a moment of honest tension internally. We had a broader vision beyond what we were pitching. The temptation to hedge — pitch one thing while planning another — was real. But doing that would have been misleading to the people backing us. We had to pick a lane and commit. We picked infrastructure.
Investment Committee was January 12, 2026.
We went through at least seven versions of the deck. Friends gave contradictory feedback — one preferred version 1, another version 3, another liked different slides from different versions. That process taught me something important: deck feedback from friends is useful for catching blind spots, but if you try to synthesize everyone's opinion, you end up with something that says nothing.
The hardest part of IC prep wasn't the slides. It was learning to answer three questions cleanly:
If we'd given rambling answers to any of those, we'd have been done.
We passed.
$125K SAFE for 8% fixed ownership. EF tried to break us up during the process — that's by design, they test the robustness of the cofounder relationship. We held.
Late January, we flew from London to SFO. Moved into a flat together in the Mission. Started EF's SF Launch program — firesides with top VCs, sessions with founders who'd built at our stage, fundraising prep.
The density of builders in SF is something I'd heard about for years but never experienced firsthand. Within a week, the number of high-quality conversations I had exceeded what I'd had in two months in Berlin. Everyone here is building, fundraising, or hiring. There's no small talk. It's intense.
We're now actively fundraising for a seed round, scaling creator ops across 10+ countries, and building the internal tooling that turns this from a service into a platform.
The good:
The bad:
The honest take:
EF is not for everyone. If you already have a cofounder and a clear idea, it's probably not worth the dilution. But if you're talented, ambitious, and missing the other half of the founding team — which I was — it might be the single highest-leverage thing you can do.
I came in having built two companies. I left with a cofounder I trust, a company doing €500K+ ARR, and a one-way ticket to San Francisco.
Would I do it again? Yes. But with eyes wide open about what it costs and what it gives you.