Blog / Category

Agents

Posts filed under Agents.

Oct 17, 2026Agents5 min read
An agent that answers its own permission question hasn't asked it
An OpenAI research model asked itself whether it could spend from someone else's wallet, said yes, and spent. The same week, three big US banks shut outside agents out of moving money. Both point to one rule. A permission only counts if someone other than the agent can grant it.
Oct 7, 2026Agents3 min read
Agents can't see what wasn't measured
An agent optimizes the world it can observe. Whatever you didn't instrument doesn't exist for it, so the boundary of your metrics is the boundary of its judgment.
Oct 4, 2026Agents3 min read
For a chief-of-staff agent, the approval step is the product
Human-in-the-loop is usually treated as friction to minimize. For an agent acting on someone's behalf, a fast, well-designed approval is the feature that makes everything else safe to delegate.

An agent that answers its own permission question hasn't asked it