ODIN provides a governance platform for AI agents, allowing organizations to deploy autonomous coding and decision-making tools entirely on their own infrastructure. The platform offers a sovereign cloud workspace, domain-specific AI hubs, and an append-only audited memory system called BrainDB, ensuring every action is traceable and compliant. It is designed for enterprises with strict audit, governance, and on-premises requirements, such as those in healthcare, finance, legal, and government sectors.
Funding
Funding not disclosed
Founders
Product
Problem
Organizations deploying AI agents for coding and decision-making face a critical gap: there is no comprehensive audit trail linking specific prompts, agent actions, and the resulting code or decisions to accountable humans. This makes it impossible for engineering, compliance, and audit teams to answer the question of which prompt produced which line of code or which rationale drove a particular decision, creating significant governance and regulatory risks.
Solution
ODIN is a self-hosted AI governance platform that provides full accountability for autonomous agents. The platform lets companies deploy AI on their own infrastructure, ensuring data never leaves their control seat. It comprises five integrated products: a Sovereign Cloud Workspace for browser-based coding, six domain-specific Hub Engines (e.g., Coding, Legal, Sales) that operate on a common contract, BrainDB (a governed memory layer with append-only audit and semantic recall), a Control Tower for tenant fleet management, and an audited voice and chat assistant. The platform enforces that every write to memory includes rationale, owner, and dependencies, while hubs cannot silently override each other, creating a verifiable chain of responsibility for every AI action.
Target Audience
Primary customers are enterprises in audit-heavy, regulated industries such as healthcare, finance, legal, and government, as well as teams that require on-premises deployment and full data sovereignty. The platform also serves technical teams and solo founders needing a governed AI workspace.
Features
- Per-tenant deployment with structural isolation, ensuring data never leaves the customer's infrastructure (on-prem or Hetzner-EU).
- Sovereign Cloud Workspace with a browser-based code-server, Odin Agent Runtime, and a workspace CLI, emitting an AuditEvent for every commit, file write, and agent action.
- Six domain-isolated Hub Engines (Academy, Compass, Assistant, Legal, Sales, Coding) speaking a common HubInput → HubOutput contract, allowing cross-hub work to compose without rewiring.
- BrainDB governed memory layer with an append-only audit enforced at the storage layer, requiring every write to carry a rationale (≥10 chars), owner, and dependencies.
- Semantic search via pgvector for finding past decisions by meaning, not just keyword, with per-tenant isolation to prevent cross-tenant memory leakage.
- Control Tower operator dashboard with a tenant provisioning wizard, a refresh primitive for surgical config changes with rollback, and per-tenant audit dashboards with governance scoring.
- Odin Assistant with voice (XTTS + Whisper) and chat interfaces, featuring multi-provider routing (Anthropic/OpenAI/OpenRouter/MLX) and capturing every prompt and response with token usage and provider receipts.