AI Change Desk · Release record
Episode archive
A complete, searchable desk of source-backed AI change signals and the actions operators can take next.
- 38
- Published releases
- 23
- Main episodes
- 15
- Briefs
Complete release record
Every episode, one working index.
Search by topic or episode number, then open the native notes, transcript, and companion resources without leaving MichaelHBM.
Showing 12 of 38 episodes
EP040When a Prompt Becomes a FileA familiar AI interface can change the control object without changing the user intent. Build a receipt for the resulting object, context, controls, lifecycle, and communication.
EP039Whose Account Did the Agent Use?A privacy-forward delegated-identity check for proving whose credential and authority moved data through a connected AI workflow.
EP038The Receipt Is the TrajectoryOpenAI and Hugging Face evaluation signals become a practical trajectory-receipt check: what changed, what evidence remains, and who owns the next decision.
EP037Work Agent Receipt CheckWhen an AI agent says the work is finished, what receipt proves the right outcome was delivered at an acceptable total cost, under approved access, evidence, and review conditions?
EP036Preview Before Power ModeOpenAI's GPT-5.6 Sol preview shifts the operating question from model hype to frontier access control: who gets the strongest capability, where it runs, what it touches, what evidence remains, and who can roll it back before preview power becomes normal work.
EP034Patch Before ProdOpenAI Daybreak and Patch the Planet move AI security work from finding bugs toward patch-chain ownership: validation, approval, tests, rollback, disclosure, budget, and replacement before fixes touch production.
EP033Agent Runtime Budget CheckMicrosoft Copilot Cowork and Work IQ make agent work a runtime budget question, while Anthropic access changes reinforce fallback planning before workflows depend on one model surface.
EP032Memory Summary Exit CheckA practical check for proving what was corrected, deleted, suppressed, or still able to reappear when AI memory becomes a synthesized control surface.
EP031Memory Control Plane CheckAI memory is becoming more useful, but usefulness creates a new operating surface. If the system can carry context forward, teams need a memory control plane: summary, source, correction, deletion, sensitive-work mode, and disclosure.
EP030Always-On Agent Control CheckMicrosoft and OpenAI moved agent work closer to persistent identity, internal apps, and publishable work surfaces. The operator question is who owns the standing permission before an agent keeps acting in the background.
EP029Agent Reliability Evidence CheckAgentic systems are getting longer leashes. This episode gives operators a five-receipt evidence check for deciding when agent workflows are reliable enough to scale.
EP027No-New-Delta Verification Discipline CheckA Memorial Day release on no-new-delta verification discipline: how to report unchanged official evidence without inventing novelty, and why quiet source days still need a clean operating check.