Reliability baseline
Keep frankai.online live, rebuild memory automatically after ingest and verify public routes after deployment.
First release cycle
Execution priorities before expansion.
The first FrankAI platform release should be judged by controlled usefulness: memory works, actions are governed and the release surface is honest.
Security boundary
Require bearer authentication for memory APIs and keep raw private payloads out of public citations.
Pilot readiness
Define owner, allowed tools, approval gates, action receipt format and rollback path for each pilot workflow.
Docs and onboarding
Publish the minimum viable platform docs so serious collaborators can understand what is live and what is staged.
Model portability
Keep memory and governance outside any single model vendor so the platform can change engines without losing continuity.
Operating loop
01
Stabilise memory ingest and search
02
Publish platform governance and rollout structure
03
Select one controlled pilot workflow
04
Instrument approvals, logs and receipts
05
Review evidence before expanding access
The release cycle is a control surface: it says what matters now and what is deliberately later.
Open docs