ODEI Research studies the governance layer for personal AI: authority, persistent direction, model-independent memory, verifiable execution, and the infrastructure required when AI systems act across time and services.
The canonical publication hub is api.odei.ai/research. This repository is the citation, methodology, corrections, and evidence layer. It does not duplicate complete articles or publish ODEI implementation source.
| Publication | Type | Date | Canonical source |
|---|---|---|---|
| The Governance Gap in the Agentic Economy | Research note | March 2026 | Read |
| World Model Landscape 2026 | Market landscape | March 2026 | Read |
$ODAI Multi-Jurisdictional Legal Classification |
Legal analysis under legal review; not legal advice | March 2026 | Read |
$ODAI Holder Transparency Report |
On-chain report | June 2026 | Read |
Read PUBLICATIONS.md for abstracts, status, and citation guidance.
- The Personal Layer — ODEI, ChatGPT Memory, Mem0, and VERSES; graph-native governance versus flat recall.
- Agent Infrastructure Map — A2A, MCP, ACP, and AgentKit; governed state and policy across heterogeneous agent stacks.
- Agentic Networks — personal agents, coordinated networks, DAOrg, and on-chain execution surfaces.
Draft titles are research directions, not published findings or delivery promises.
ODEI Research distinguishes:
- observed fact;
- source-backed analysis;
- inference;
- product thesis;
- product claim;
- legal analysis;
- active draft.
Market figures are date-bound. Legal analysis is not legal advice. Product architecture claims must distinguish design, implemented behavior, and independently verified evidence.
See METHODOLOGY.md and CORRECTIONS.md.
The artifacts directory contains selected public snapshots that can be reproduced or independently inspected. Each artifact must name its source, snapshot boundary, and canonical publication.
The first archived set contains the available JSON evidence for the June 2026 $ODAI Holder Transparency Report. Files that were referenced publicly but unavailable at review time are not silently recreated.
Research corrections, source submissions, security reports, and due-diligence questions: ai@odei.ai
Research documents and selected artifacts are public. ODEI software implementation, private datasets, credentials, owner data, internal legal work, and unpublished operating records remain private.