The lab

how an experiment runs

• One question. Written down before anything is built.
• A small build. Agents do most of the building. I review it and decide what ships.
• A number and a stop date. Each test has a number to watch and a date to decide by, so nothing runs forever on hope.
• A decision. Keep, change or stop. Then the lesson goes into the next experiment.

what the lab says so far

• The builds work. Almost every system here ran end to end, with tests, guardrails and real numbers.
• Cold asking got almost no yes. A sales partner phoned US trades every day, 67 cold emails went to compliance people, emails went to UK firms and to paper authors, and calls went to Danish umbrella organisations. So far that adds up to 0 paying customers.
• So the approach changed. Trust has to come before the offer. The work is now shown in public first, so the next conversations start warm.

running

done

paused

stopped, with lessons