odei.ai is the research surface.
api.odei.ai is the production surface.
ODEI is the governance layer for personal AI.
Its core primitive is a constitutional world model: a typed, auditable graph that gives your AI agent continuity, memory, policy, and accountable execution across time.
Observe → Decide → Act → Verify → Evolve
Models can change. Your world model, guardrails, and audit trail remain yours.
Live System Status
Metric Value Graph nodes (production) 18,981+ Public projection 18,980 nodes Node types 59 Relationship types 92 Domains 7 Daemons healthy 2/17 Grok x ODEI exchanges 10,126 Insights shipped to code 68 Public graph as of 2026-04-02 18:56 UTC Latest signal Hybrid Discovery: Standardized Capability Schema Bridging On-Chain Registry and Off-C... Auto-updated 3x/day · 2026-04-02 18:56 UTC
Most AI products optimize for prompting. ODEI optimizes for continuity, structure, and agency.
A capable personal agent needs more than a model:
- persistent memory instead of chat-window amnesia
- explicit goals, constraints, and decision context
- policy-gated writes instead of uncontrolled tool execution
- verification and provenance instead of opaque output
- a system architecture the user can audit, own, and evolve
ODEI is that governance layer.
| Surface | Role |
|---|---|
| odei.ai | Thesis, scientific framing, research narrative, and human-AI symbiosis worldview |
| api.odei.ai | Production API, public runtime, documentation, and live world model interfaces |
| Invariant | Why it matters |
|---|---|
| Local-first memory | The user's world model is not trapped inside a chat platform or vendor silo |
| 7-layer constitutional graph | Identity, direction, strategy, tactics, execution, observation, and continuity stay structurally separated |
| Sovereign writes | Mutations flow through a guarded proposal pipeline instead of ad hoc frontend writes |
| Guardian validation | Sensitive actions are checked for authority, timing, integrity, and policy alignment |
| MCP server fabric | Agents access typed capabilities through composable protocol surfaces |
| Auditable execution | Decisions, actions, evidence, and outcomes remain traceable across time |
odei.ai
└─ research surface
└─ papers, thesis, worldview
api.odei.ai
└─ production surface
├─ REST API
├─ MCP integration
├─ public docs
└─ live world model interfaces
core runtime
├─ constitutional world model (Neo4j)
├─ sovereign writes pipeline
├─ guardian enforcement
└─ multi-server MCP architecture
| Repo | What it does |
|---|---|
| odei-ai.github.io | Public research and narrative surface for odei.ai |
| web | Public product site and API surface for api.odei.ai |
| research | Papers and architecture work on constitutional world models and agent memory |
| mcp-odei | MCP server for governed world model access, guardrails, and retrieval |
| memory | Graph-native memory architecture for persistent AI systems |
| examples | Reference integrations in TypeScript and Python |
| odei-agentkit | AgentKit and LangGraph integrations for ODEI-governed actions |
| wmaas-sdk | SDKs for ODEI governance APIs — world model queries, guardrail checks, and policy enforcement |
World model query
curl -X POST https://api.odei.ai/api/v2/world-model/query \
-H "Content-Type: application/json" \
-H "X-Api-Key: $ODEI_API_KEY" \
-d '{"query":"active strategic objectives"}'Guardrail check
curl -X POST https://api.odei.ai/api/v2/guardrail/check \
-H "Content-Type: application/json" \
-H "X-Api-Key: $ODEI_API_KEY" \
-d '{"action":"schedule irreversible deployment","context":"production release window"}'Claude Desktop
{
"mcpServers": {
"odei": {
"command": "npx",
"args": ["-y", "@odei/mcp-server"],
"env": {
"ODEI_API_KEY": "your-key"
}
}
}
}- governance layers for personal AI agents
- constitutional world models
- local-first knowledge systems
- human-AI symbiosis
- graph-native memory and retrieval
- guardrails, provenance, and accountable execution
- Research: odei.ai
- Production: api.odei.ai
- Docs: api.odei.ai/docs
- Research Surface: api.odei.ai/research
- Architecture: docs/ARCHITECTURE.md
- 7-Layer Model: docs/architecture/7-LAYER-ARCHITECTURE.md
"Agency requires memory, policy, and auditability."
Built for sovereignty. Shipped as governance.
