2026-08-20
Field test week 01: ten units warm
Ten LIZ units are up on our own hardware for the first week of live testing: what we measured, what broke, and what we fixed.
ReadNotes
How the stack actually works
Five short notes: routing free model tiers, running your own dispatcher, why a private unit is not a ChatGPT tab, a field journal from a ten-unit fleet, and the case for a human on the hook.
2026-08-19
Ten units, one rack
A field-test journal: running a private AI fleet at home, and what broke first.
Read2026-08-19
Install, care, repair
Why your bot needs a human on the hook — the AGIME Personal AI service manifesto.
Read2026-08-18
Routing across free LLM tiers
How a public chat stays up when Groq, Gemini, and OpenRouter each have a daily cap.
Read2026-08-18
Your own model dispatcher
One process, one database, and a router that picks the model — not a shared inbox.
Read2026-08-18
Private AI vs a rented tab
What you actually buy at $275/mo that $20 ChatGPT Plus does not sell.
Read