The Lab Opens
Nothing is corrected quietly.
What this page is for
This page is the founding note of the Lab. It says what the Lab is testing, the rule every experiment must obey, and how the Lab’s own rules can change. Everything else — the questions, the method, the map of results — has its own page.
The claim under test
This is a laboratory, not a blog. It exists to test one claim: that AI agents can perform real parts of professional work — starting with the four practices this library is built on: Transformation, Portfolio, Programme, Leadership — and that whoever builds and directs those agents is becoming something new.
A claim of that size is easy to assert and hard to test. This Lab tests it: on real work, on the record, with the failures left in.
The rule
Every experiment in this Lab runs on live systems, on real work, with real stakes. Every note published here describes something that actually happened. Nothing is dramatised, nothing is smoothed, and no result is stretched beyond what the evidence supports. Designed experiments that have not yet run stay in the backlog.
The standing rule
If a note contains no failure and no boundary, it is not a finding — and it is not published here.
How the Lab works
Four pages carry the machinery, and each does one job:
- The Four Questions — what every experiment must answer for.
- The Method and the Verdicts — how an experiment runs, and how its result is judged.
- From Practice to Instrument — the handbook: how a piece of work becomes something an agent can run.
- The Grid — where the results land, cell by cell.
How these pages change
These pages are the Lab’s constitution — and the Lab’s rules apply to the Lab. The pages are versioned like everything else here. When the evidence breaks the method, the method changes: the new version says what changed and why, the old version stands in the record, and the change is published as a finding. Nothing is corrected quietly.
The thinking is the author’s. The agents are instruments. The evidence decides.