AI at work

A meeting note-taker that records and summarises without joining as a bot.

A high-volume inference consumer paying retail — every meeting they capture is transcription plus summarisation billed per token, and their margin is the gap between a $14–$39 seat and what that seat costs them to serve.

AI agents that run customer support and back-office work for financial firms.

The applied end of the agent stack — what it takes to run an autonomous agent inside a regulated bank, where the eval harness and the audit trail matter more than the model.

An “AI workforce” for UK small businesses that reads their inbox, calls and CRM.

The applications end of the stack — they sell the outcome of inference to non-technical UK businesses, so their pricing is where token cost becomes a task price a buyer actually sees.

An AI coworker that watches you do a task, then offers to take it over.

The clearest published pivot in the agent market — a $30M seed spent building developer infrastructure, then turned around in public to build the end product instead, with the whole argument written down and dated.