What we got wrong about vertical AI in 2024
We passed on four companies for the same reason, and three of them are now doing well. Here is the mistake we made and the question we ask instead.
Insights
Notes from the work, written by the people doing it.
We publish when we have learned something, not on a schedule. Memos, field notes, and the occasional guide we wish had existed when we were operating.
Latest
What we got wrong about vertical AI in 2024
We passed on four companies for the same reason, and three of them are now doing well. Here is the mistake we made and the question we ask instead.
We had a rule that vertical AI companies needed proprietary data to defend a position. It turned out that distribution into an unloved workflow was worth more, at least at this stage of the cycle.
The question we ask now is simpler: who is doing this work today, how much are they paid, and what would make them switch on a Tuesday morning.
Archive
Everything we have published.
Pricing when your costs move every quarter
Inference costs fall, usage climbs, and the pricing page you wrote at launch stops making sense. Notes from eleven portfolio companies that repriced.
The first ten hires still decide the company
Our talent team looked at hiring data across the portfolio. The pattern is unglamorous: teams that hired slowly at the start moved faster at Series B.
A seed round data room that takes two hours to build
Most data rooms are either thin or bloated. This is the version we ask for, with the reasoning behind each item.
Selling to utilities without losing a year
Tobias spent a decade inside one. He wrote down how procurement actually works, including who can say no and who can only say yes.
Why we still lead pre-seed rounds
Ownership matters less to us than being early enough to be useful. An argument for writing the first cheque, including the part where it hurts.
Board meetings that founders do not dread
A short format we use with early companies. Two pages, one decision, no theatre.
What happens after a down round
Three founders in our portfolio agreed to talk candidly about resetting valuation, telling the team, and getting back to work.
Formats
Three kinds of writing.
Memos
Investment thinking, including the times we were wrong. Usually written by whoever led the deal.
Field notes
Short pieces from inside the portfolio. Pricing changes, hiring data, sales cycles that took longer than planned.
Guides
Practical documents founders can copy. Data room checklists, board formats, reference call scripts.
Tell us what you are building.
Send a short note. We read every one and reply within five working days, whatever the answer.