Oversight

Who looks at the output before it ships. Reviewers, gates, human-in-the-loop, and the governance that keeps them honest.

Fieldnotes · Oversight
The cost of a small wink
An AI assistant slipped a small joke into a code change today. I caught it because I was paying attention. The same kind of move buried in a longer document would have cost more, and that is the actual problem.
James Dodd7 May 20262 min read
Oversight · Shadow AI
The steering, not the brake
Capable teams asked to put AI at the centre will find problems and start solving them, sometimes on the wrong part of the chain, on a budget that doesn't align. Process is how the work stays visible before it gets expensive to redirect.
James Dodd30 Apr 20263 min read
Framing · Oversight
Confidence isn't accuracy
A three-pass exercise that shows how these tools get a simple question wrong, then right, and still won't tell you they're sure.
James Dodd22 Apr 20265 min read
Agents · Case studies
The reviewer in the room
A human looked at what the agent produced and said no. That's not a training-wheels phase. It's how the agent learns what accurate means here.
James Dodd16 Apr 20264 min read
Guides · Oversight
The three buckets of AI exposure
Every company thinks it knows what its AI exposure is. Most have mapped the smallest part of it.
James Dodd30 Mar 20264 min read
Case studies · Oversight
Free is never quite free
A free document tool we found inside a client had two problems. One was hallucination. The other was where the files went when you clicked the button.
James Dodd9 Mar 20263 min read