/ 01
Ouroboros
adaptive, model-agnostic workspace · switch models, organize projects & carry user preferences across work
user-owned context as infrastructure for the next generation of agentic systems.
sierra catalina builds privacy-first systems for user-owned intelligence, portable context & cooperative automation.
01thesis
the missing layer in modern AI is portable, user-owned context. memory, permissions & identity should travel with the person across models, agents & applications. context is becoming infrastructure. it should belong to the user, not the platform.
02operating principles
- AI systems need scoped access to context, not persistent ownership of identity.
- provenance, revocation & portability are baseline requirements for trustworthy personalization.
- the future of AI is hybrid.local, cloud & agent systems will coordinate through portable context layers.
03current work
live / in build
Ouroboros · a model-agnostic AI workspace for switching models, organizing work & carrying memory and preferences across projects. its customizable interface is designed for intelligent routing & is being built on the context layer protocol.
04work
/ 02
proof of work
xAI, prontoAI*, pollen mobile, hotspotty, helium
/ 03 · open source
agent-aware architecture
open reference starter · explicit intent, machine-readable discovery & agent-friendly HTTP surfaces
/ 04 · protocol
context layer
user-owned context protocol · purpose-bound disclosure, reversible writes & inspectable receipts