I always wanted to build robots.
The dream didn't last long. Robots were 'pretend.' Not a real job. Definitely not work for folks in Burnley. Still, I dreamed.
During my MSc I learned how hard robots are. Temperamental. Mathematical. Problematic. I shelved the dream and spent a decade building B2B startups.
But then ChatGPT happened.
Vision models spot tumours. Language models write essays. The golden path to real machines suddenly comes into focus. Ilya, then at OpenAI, calls shelving robotics a mistake. Elon unveils Optimus. Figure has robots used in factories.
Robots are coming. But aren't they missing the obvious jobs?
Take asparagus. At dawn a worker walks each row, underpaid and overworked, eyes the spears, and slices those at the right height. Labour is forty per cent of the price. And that's before you count the human cost.
But Optimus will never plough a muddy field. It's not going to work in rain or sleet. And what a waste if we did! All those servos in the hand to cut an asparagus? Put the same brain in a low, cheap, rugged rover built only to cut spears.
What if we could make farming profitable again? Mixed crops in the same acre, each plant tended by tireless bots. Machines for specific tasks.
That's why we started adam. Paul Graham says sprint to the frontier and do great work there. Our frontier is wet, muddy, and ready.
So, in short, the master plan is:
Don't tell anyone.