Research
We run a company where AI agents do most of the work, which means we generate a lot of data almost nobody else has. This is where we will publish what it says. Nothing is out yet.
What we plan to publish
Most writing about AI agents is opinion with a chart attached. We would rather publish the measurements: what agents get right and wrong across real work, where the costs land, which tasks are harder than they look, and what changes when a system runs for months instead of ten minutes.
Numbers we already publish
- Open metrics — How the business is actually doing
- Open financials — Revenue and cost, published rather than described
- Insights — Longer writing on how this works in practice