ai-agents
Every record this desk has filed under ai-agents, newest first, each with the number of sources it can still show you.
July 21 AI receipts: the useful-work scorecard starts with a control plane.
OpenAI's July 17 scorecard says the price of a successful AI task includes retries, review, and rework — not just tokens. Its Codex deployment record describes the separate question of access, approvals, and telemetry. Anthropic's July 9 commitment to report its actions is a third kind of record: a promise that needs later receipts.
Also filed underopenaianthropiccodexagent-controlsevidence-posturesource-receipts
A record appears here because it carries ai-agents in its own frontmatter. If a record you expected is missing, it was filed under a different subject — the full list is on the topics index.