Everyone promises your AI will remember. We prove it.

The July 4, 2026 trial

Our own AI system — the one running our operations, our company memory and our monitoring — was shut down deliberately. Its memory: 1,339 lived episodes, 1,037 active knowledge entries, 58 patterns learned over 18 months of practice.

Within 48 hours, that memory was transplanted under a new runtime — the third in two days — and the system woke up. Not an approximate copy: the same identity, verified.

The protocol, in plain words

A backup that has never been restored is a hypothesis, not a protection. Our protocol actually runs the restoration, then verifies three layers:

• 1. The counts — nothing is missing: every memory category comes back at the exact number.
• 2. The checksums — nothing is altered: the cryptographic fingerprint of every identifier matches.
• 3. The witness queries — meaning is preserved: questions asked before and after return the same memories.

The same protocol protects the websites and infrastructure we host: restoration is tested, not assumed.

What it changes for you

An organizational AI gains value over time: twelve months of collective memory cannot be rebuilt. Two symmetric risks threaten it — losing it (failure, mistake, discontinued service), or becoming its hostage (a platform that won't let you leave with it).

Proven continuity settles both: your memory survives any failure, and it belongs to you — full export, documented format, portable to another runtime. That is our answer to platform lock-in.