4-Layer Architecture
A layered approach to AI orchestration that separates strategic intent, routing, execution, and truth storage.
Executive Layer
Strategic intent, decision authority, scope control
Hermes operates as a context-aware operator surface. MCP is the read-only bridge for context retrieval — not an autonomous execution path.
Routing Layer
Capability-based routing, worker allocation
Execution Layer
AI workers, policy guardrails, testing & critique loops
Human Review Gate
Checkpoint, not a layer — every artifact passes here before it is treated as committed truth.
Source-of-Truth Layer
Committed KB (glossary + index), decision logs
*Current operational trial surface. MCP is used as a read-only context bridge for controlled context retrieval. It is not yet approved as an autonomous production workflow.
Cross-cutting: Backlog & Visibility surface — control surface across every layer
AIOS Observability Layer
Captures execution traces, routing decisions, cost/usage evidence, reliability signals, and deployment readiness checks. In Phase 1, telemetry production and store behavior remain implemented in optimize-worker while AIOS owns the governance meaning of the evidence.
Surface
ai-os-profile remains the deployable cockpit and app surface.
Foundation
AWS foundation is ready, but the app is not deployed there.
Gate
Deployment waits for the SQLite runtime contract; refactor work is paused pending regression strategy.
Six Design Choices
Match capability to task
Boundaries before behavior
Control surface, not admin
Truth lives in committed knowledge
Critique, regression, gates at every layer
Last line before truth commits
“AI orchestration is program management. Tools change. Governance discipline doesn't.”
Component Glossary
| Component | Definition |
|---|---|
| Robert KB + Git | Source of truth. Master archive of decisions. |
| Robert / GPT | Executive orchestrator and reviewer. |
| Hermes + Opus + custom MCP* | Context-aware operator surface — Active trial — structured context pull, task continuity, governance-aware operation. |
| optimize-worker | Execution routing and task packaging layer. In Phase 1, it also implements telemetry production and store behavior for observability evidence. |
| Codex | Bounded implementation executor for repo edits. |
| ai-os-profile | Curated external visibility layer and deployable cockpit/app surface (this site). |
| AIOS Observability Layer | Governance evidence layer for traces, routing decisions, cost/usage evidence, reliability signals, and deployment readiness checks. Not a claim that AWS app deployment is complete. |
*Current operational trial surface. MCP is used as a read-only context bridge for controlled context retrieval. It is not yet approved as an autonomous production workflow.