Description
Olivares AI is an open-core control plane for enterprise AI that runs entirely on your own infrastructure. It gives platform and security teams one place to see and govern the AI agents, sessions, models, and MCP servers already operating across their estate, spanning Claude, Codex, OpenAI, Gemini, and any other provider or model you host yourself.
Read More
Deployed as a single container or binary, Olivares AI passively discovers running agents from a minimally privileged, read-only collector, then builds an access graph that maps what each agent can reach versus what it actually touches, distinguishing read from read/write. From there, teams set policy, watch for least-privilege drift, and export audit evidence, with guardrails enforced at access time rather than only logged after the fact.
The platform is designed to run air-gapped, making no mandatory calls home, and keeps governance data under the customer's control. Security features include mutual TLS on every internal hop and an append-only, hash-chained ledger for tamper-evident records. It also tracks coverage against frameworks such as the EU AI Act, NIST AI RMF, ISO 42001, and SOC 2.
Olivares AI is released under AGPL-3.0 with the complete product published to a public repository, and a commercial license plus a dedicated enterprise tier are available for organizations that need them.
- CategoryMLOps & Model Deployment
- Pricing ModelOpen source
- DeploymentSelf-hosted
- PlatformsLinux, Windows, macOS
Olivares AI's Core Features
Self-hosted, vendor-neutral control plane for enterprise AI
Passive discovery of agents, sessions, models, and MCP servers
Access graph mapping permitted versus observed read/write access
Policy enforcement and least-privilege drift detection at access time
Audit evidence export with an append-only, hash-chained ledger
Mutual TLS on every internal component hop
Air-gapped friendly with no mandatory calls home
Compliance coverage tracking for EU AI Act, NIST AI RMF, ISO 42001, and SOC 2
How to use Olivares AI?
Deploy: Run a single container or binary on your own infrastructure, with no account or cloud dependency required.
Discover: Let Olivares AI passively find every agent, session, MCP, and model already running.
Map: Review the access graph showing what each agent can reach and what it actually touches.
Govern: Set policy, watch for drift, and export audit evidence as your AI estate grows.
Olivares AI's Use Cases
- AI agent visibility
- Least-privilege enforcement
- Audit and compliance
- Air-gapped deployments
