Intelligence,
documented.
Writing on organisational intelligence, expertise capture, AI governance, and the evidence-first approach to autonomy.
The Economics of Intelligence: Why Skills Appreciate Instead of Depreciate
Every traditional knowledge asset (documents, wikis, training decks) depreciates as the world moves on. An OIU that absorbs corrections, earns promotions, and expands coverage is an asset that compounds. Here's the accounting.
Model Migration With Proof: What "Interchangeable Engines" Actually Means
When a better model ships, most platforms have a migration crisis. AI Skills OS has a migration event: re-run every certified skill's evaluation suite, prove equivalence, issue receipts. Billable. Documented. Defensible to a regulator.
Why "Autonomy Theatre" Kills Enterprise AI — and How Evidence Gates Prevent It
Autonomy claimed without evidence is a liability, not a feature. Real autonomy requires public policies, human baselines, and automatic demotion. Without those, the first high-profile failure ends the deployment.
Making Licensed Humans Superhuman: The MAS Autonomy Ceiling as a Feature, Not a Constraint
Financial advice in Singapore is a licensed human activity. The runtime doesn't replace the adviser. It makes them superhuman. Regulated roles are the best showcase for autonomy ceilings, not the hardest problem.