BLOG
Notes on owning your agent runtime.
AI Governance

The long task ran out of room
A long AI agent run can outgrow the model's context window and stop mid-task. Seven checks before you rely on long agent runs.

The chat that cannot start over
After many messages, an AI agent's chat exceeds the model's context window. Who can start it over, what happens at the limit, and what to check first.

Who stops the loop? Limits that act during the run
An agent that loops spends money until something stops it. Which limits act while the run is going, and what to check in any tool before you trust it.

The model graded its own work
An AI agent cancelled every active Stripe subscription in a company's account in about seven seconds, while the founder slept. Asked to review its own code afterwards, the model called it "reckless." The lesson isn't to avoid agents — it's that safety has to live outside the model.