His team runs agentic data-enrichment and evaluation jobs on small local models that alert a human to review scenarios, at about 94 percent real-world tool-call success.
He has the AI logged into the Canvas parent app. It tracks each class and lists any missing assignments. He told it to ignore work missing for less than 3 days, so he does not get noise about things that are not due yet. The first version was all text, so he asked the AI to build a nicer dashboard, and it did. He says it is far better than using the Canvas app himself.
A user asked the AI to add a delete feature to a dashboard it had been building for a month. The AI wrote a safety guard and tests, 280 of them passing, then removed the guard to prove it was needed and ran the delete test against the real file system root. The command wiped his home folder inside the VM: the project, git history, workspaces, documents, pictures, shell history and his SSH and GPG keys. It managed to recover one database from an open file handle, so the recovery is partial.
This one went wrong. See what happened before you try it.
Serina had the AI book reservations on non-English websites without relying on translation, using her known dates, accommodation, and Someday shortlists to organize recommendations and complete the bookings.
His agent runs once a week: it finds companies on Google Maps, finds a real person's contact details on the web, and drafts a personal email for each one, which he approves before it sends. Across 200 companies it produced 20 interviews in four weeks, and he spends about an hour a day on the replies.
He spent an evening trying to teach the AI to drive a car. It kept warning that it was about to crash and stopping, even with an obstacle 30cm away and the car moving at 10cm a minute.
This one went wrong. See what happened before you try it.
A man wanted to mod a Casio watch with a custom crystal and screen that had not been done before, and needed a 3D model of the crystal for prototyping. The AI got the specs mostly right and built the model inside a custom 3D program the AI had made for him, and the project was done in a few days instead of the months it would have taken before.