Lab
What I build, what I test, and how you can do the same.
Experiments
Backups: encrypted notes every ten minutes, code every night
Three small jobs copy everything that matters: my notes, my code, and the family app's files. When a backup falls behind or fails, I find out.
A Google Drive that tidies itself every month
I sorted out my Google Drive once with AI, then set a monthly task to keep it that way: it files new documents, renames them, removes duplicates and tells me what's about to expire.
The drop folder: I drop a file, AI files it
One folder on my computer takes any file. AI files it into my notes three times a day, I get pings, and the folder tidies itself every week.
Email and texts to tasks: appointments don't get lost
Texts about appointments are forwarded to my email, and twice a day AI reads my email and turns what needs me into TickTick tasks. They then show up in my morning brief and in Signal.
Supermarket offers every morning, politely
Every morning, a job collects supermarket offers for an app I built for the family. The pages give the prices; AI reads the product pictures. And the job stays polite: limits, gaps between requests, and it stops when a site says no.
A daily homelab check that reads, proposes and touches nothing
Every day at 9, AI checks the server at home: disks, apps, logs and backups. It reports what it finds and proposes fixes; approved repairs happen separately.
Innovation radar: five agents search, one checks
Every Sunday, AI looks for what changed in my field that week and gives me the few things I can actually act on. Several models search side by side, and one checks everything.
Machine health and the guard: a check every morning, and an eye on what goes public
Two read-only jobs: one checks my always-on computer every day, the other flags unexpected changes in the settings that publish my sites.
Morning brief: what needs me today, in a few cards
Every day at 7:30, AI reads my email, calendar, tasks and notes, and gives me a few cards on what needs me. I answer each one with a tap.
The school app: notices and dates, neatly in order
Twice a day, a read-only script connects to the app the school uses with my own login, pulls the new announcements and dates, and sorts them into files. Then AI turns what needs me into tasks.
Skill updates every Sunday: update, test, roll back what breaks
Every Sunday morning, an agent in T3 updates every Skill and tool I use to its latest version, tests it, and rolls back anything that breaks.
The start page: every home tool on one screen
A page whose data refreshes every five minutes, showing whether each home tool is up or down and since when, the scheduled jobs and their results, and the server's storage and internet speed.
T3: every AI model I use, in one place
Claude lived in one app and Codex in another, each with its own limits. Now they all run in T3, work together, and I can switch between them any time.
