Experiments

Testing before deciding. Benchmarks, comparisons, and the small trials that answer big questions.

Framing · Fieldnotes
The Waitrose effect, but in your data
Free public data is sitting there, ignored, because on its own it looks like noise. Joined to what you already know, it points somewhere. The join is the part worth paying for.
James Dodd25 Jun 20263 min read
Agents · Case studies
We built one AI to commit fraud and another to stop it
Both agents found a way. Neither was broken. That was the problem, and it taught us where AI earns its keep and where it doesn't.
James Dodd17 Mar 20267 min read