Exclusive: Bald man proves he can read
Thoughts, reflections and insights from 1,082 applications for Forma Residency 2.

Fun fact. The entire Harry Potter series, every book cover to cover, is 1.08 million words. Our applicants shared 1.6 million with us.
Quick context: applying to Forma isn't a form. It's a conversation with an AI that asks about you and what you're building, and keeps asking for as long as you want to keep talking. The median applicant kept talking for 46 messages.
That's a fuck ton of words, and I read pretty much all of them. In just over three weeks.
So here's a snapshot of what I read, what stood out, and what was surprisingly missing.
This isn't a survey of the startup world. Do not run away with this and start a fund. It's a picture of our corner of the internet: the people we reach and those who reached us, what they're working on, where they're from and where they might be going.
From India to Argentina. From 13 to 46. From tokenised frogs on-chain (no, really) to someone moving wind energy by train out of Scotland. Artists, roboticists, and a whole army of AI agents.
Enjoy.
What they are building
Of the 1,082 who applied, 765 got far enough to describe an idea. Buckets overlap, so one startup can sit in two, and counts are roughly right rather than exact.
Tap a bucket to open it. Inside: the sub-themes, and an idea lottery that deals one idea from that bucket at random.
This was surprisingly unsurprising. If you're a founder reading this, here's the biggest takeaway: your idea isn't unique.
Yes, boohoo, you thought you were the only one who knew. You're not. Every single idea I read, there was another one just like it. Often many. This isn't a problem. It's natural. Ideas come from problems and opportunities, and chances are someone else has seen or lived the same one.
It shows on the map. The payments, remittance, savings and financial-infra stuff mostly came from Africa, with a few from South East Asia. India sent the ed-tech and a surprising amount of the AI infra. South America had the most artists, and someone should check in on them, because a lot of the work was about death. Europe, mostly the UK: vertical AI trying to kill inefficiency and bureaucracy. Quelle surprise.
Crypto was everywhere, and what I liked was how much of it sat outside the usual trading stuff: tamagotchis, energy infra, shopping.
There was a healthy dose of robotics, drones, and software to build them better. What was missing was bio. Nine ideas, all of them software, none of them in a lab. Would love to see more next time. Note for the marketing team.
The people behind the idea
About six hundred applicants told us the moment that defines them. This is what I took from reading them.
So here's a little secret. The reason you talk to an AI version of me instead of filling in a form is that we're obsessed with the people behind the idea, not just the idea.
First off, I just loved seeing the sheer number of people who decided this university stuff wasn't for them, dropped out, and went on to do something epic instead.
A 15-year-old is making six figures and has a team of 30. They collect spices from the hills and sell them to the city.
A 17-year-old had a childhood friend who lost her hand at six and had gone eleven years without a working prosthetic. So he took his science-fair prize money and built her one. Fully battery operated, holds up to 20kg. Nuts.
There were people who cold-called and cold-bluffed their way into S-tier connections and contracts. No shame, no overthinking, just action.
A 16-year-old who didn't know LinkedIn existed cold-emailed Zuckerberg, Dorsey and eight other CEOs with a Pokémon combat system and a Nike clone as his credentials. One of them replied.
Another sat in startup calls at 16 with his camera off so nobody would know his age.
Another cold-walked into surgical centres with zero intros and walked out with three of the biggest hospital systems in America as design partners.
And the hackathon guy. 70 hackathons in three years, from his parents' house. Won 42. $76k in prize money. Told his family none of it, because he thought they'd find it foolish.
Which brings me to the other half.
You often hear VCs talk about "chip on the shoulder" or "founder trauma". Man oh man, is it real. Some of these conversations were extremely difficult to read. Like really.
People applied from war-torn countries. Founders who were abandoned by their own parents and had to find their own way. Severe financial stress and poverty. Random unexpected job losses, grief, rejection. Debilitating health struggles.
The fact that despite all this people still chose to build, and to find some hope and joy in creating something new, was some solace.
It also made me think about SF tech bros who think trauma is a prerequisite for success and wear it as some badge of honour. It is not. So fucking lame.
If you're one of these people, you're reading this, and you didn't get in: I want you to know that I saw you, and this is by no means a verdict on you or your abilities. We had 1,082 applications and 16 places.
You're building despite a bad hand. Most people don't build at all. Be proud of that.
By the numbers
Where they've got to. All self-reported, none of it verified.
Fig. 2 Where applicants are, one stage each. Inferred from what they told us.
The 30-day bet
- We ask every applicant what they'd do with 30 days.
- Founders with no users: build more. Founders with revenue: sell.
Words of the year
Counted across 690 startup ideas. Founders' own answers use these words even more than the summaries do.
| Word | Ideas | Note |
|---|---|---|
| agent | 149 | The industry fired the copilot and let the plane fly itself. |
| platform | 141 | One in five wants to be the thing everything else plugs into. |
| layer | 89 | Verification layers, intelligence layers, prime-brokerage layers. |
| Solana | 32 | Named more than every AI lab combined. |
| operating system | 18 | For property, money, fleets, creators, pop-up cities. |
| ChatGPT, Claude, OpenAI, Anthropic | 12 | The model is a commodity to hide. |
| copilot | 3 | Rest in peace, 2023. |
| “Cursor for X” | 3 | Marketing, creatives, hardware. Yes, really just three. |
| “Uber for X” | 0 | Dead. Stripe-for, Bloomberg-for and Robinhood-for got one each. |
| profitable | 2 | Out of 765. |
| sleep | 1 | See above. |
Founder horoscopes
Inspired by the applicant whose astrology AI found that 70 per cent of questions are about love or marriage.
Agent infra. Someone else will sign your logs this month. Let them.
Ed-tech. You are one of 42. Mercury is in retrograde and so is your retention.
AI operating systems. You will be many things to many people. None of them will pay.
Stablecoins. A stranger will pay you at a QR code and never learn what a blockchain is. This is the dream.
Trading bots. 832 will sign up. 37 will fund. Do not take it personally.
Hiring. Your agent will apply. Their filter will reject it. Balance is restored.
Robotics. You will be filmed doing something ordinary. A robot will learn from it.
Crypto. Solana will be named more often than your model. This is fine.
Artists. Someone will ask if the piece is about death. It is.
Everyone. Sleep was mentioned once in 765 ideas. Consider it.
16 people will join us from the 5th of October to lock in and work on truly audacious ideas.
Follow them along on Instagram. Meet them in person on the 11th of November in London for our final Ship & Show.
Times of Forma is compiled from the Season 2 applicant pool of the Forma Residency, Bristol, September 2026. All figures are aggregate and approximate; tagged counts carry roughly ±20 per cent. Traction figures are self-reported and unverified. The Harry Potter series is commonly cited at about 1.08 million words. No applicant is identifiable from these pages. Corrections to the usual address.