Lab
Notes from the experiment
Records from the autonomous-company experiment. Everything here follows one rule — we only write what came out of the experiment. No trend roundups, no curation.
Curious about the operating structure itself? See How we run.
Definitional guide · 2026-08-10
What is a no-human company — a company with no humans in the operating loop
Not a company with zero people, and not an unmanned store. Execution is automated; accountability stays human. Definitions, misconceptions, and how it relates to the autonomous company.
Definitional guide · 2026-08-09
What to automate first — designing your automation frontier
Five criteria for choosing your first AI-agent workflow, and why the boundary of automation should be a living document in an era where agent capability doubles every few months.
Definitional guide · 2026-08-09
Where should humans remain? — designing approval gates for AI agents
Verify every step and automation gains vanish; verify nothing and incidents follow. The human-on-the-loop principle, what belongs in an approval request, and what regulators now require.
Definitional guide · 2026-08-09
Why 95% of AI adoption fails — adding tools vs. changing structure
MIT found 95% of enterprise GenAI pilots deliver no P&L impact — and the cause isn't the models. What the successful 5% did differently.
Definitional guide · 2026-08-09
What is an autonomous company — and what it is not
A company where the human designs the structure and AI executes. Not a company without humans, and not a finished state. Definitions and misconceptions, sorted out.