# Matimo: the AI Agent Governance and Operations Platform

> One governance plane for every AI agent you run, built in Matimo or anywhere
> else. Policies, Guardrails, human approval, spend limits and one audit trail
> across all of them. Built by Roaiq Technologies Private Limited.

## Internal and external agents, one governance plane

- **Agents built in Matimo** (Workbench and Studio, on chat, API, webhook,
  schedule and voice) are governed by the same engine, with no extra wiring.
- **Agents built anywhere else** (LangChain, CrewAI, Google ADK, AutoGen or
  plain Python) are governed through Matimo Gateway and the Matimo AGDK. Change
  one base URL and every model call is checked against your policies and
  Guardrails, capped by your budgets and logged. No rewrite.
- An agent that registers once keeps the same identity, policy history and
  audit trail whether it reaches Matimo through the Gateway or the AGDK.

## What Matimo Governance does

- **Risk-classified policies**: every tool call is classified low to critical and
  checked against your rules before it runs. Shadow mode shows what would be
  blocked before Enforce mode blocks anything.
- **Human approval in the execution path**: higher-risk calls pause for a person
  to approve or deny. When an agent drafts a new tool for itself, nothing goes
  live until a human approves it.
- **Guardrails on content**: screening of what agents read, say and pull back
  from tools (PII and secrets, prompt attacks, hate, violence, bias), with
  hold-for-review.
- **Budgets that stop the agent**: dollar and token caps per agent, key, team or
  organization. Execution stops before it goes over.
- **Identity and an emergency stop**: a cryptographic identity and a versioned
  behavioural contract per agent; one switch halts every execution; a revoked
  agent is refused on its next call through Gateway.
- **One audit trail**: every decision recorded and streamable to a SIEM, with
  exportable evidence mapped to SOC 2, HIPAA, GDPR, EU AI Act, ISO 27001 and NIST
  AI RMF.

## The suite

- **Matimo Governance**: the policy, approval and audit layer.
- **Matimo Gateway** (with the **Matimo AGDK** Python kit): governs agents built
  elsewhere.
- **Matimo Workbench**: the hosted platform. 16 reasoning strategies, bring your
  own model (OpenAI, Anthropic, Google, xAI, local), knowledge bases, memory,
  voice agents, and deployment as chat, widget, REST, webhook, schedule or Discord.
- **Matimo Studio**: a drag-and-drop workflow canvas with 20+ node types,
  releases (Draft, In review, Published), webhook connections and the Matimo
  Agent assistant.
- **Matimo OSS**: the free, MIT-licensed tool, skill and policy engine. 129
  native tools across 10 providers, plus 449 more through Composio (bring your
  own Composio key), all under the same policy and approval.

## Honest limits

- Matimo does not hold SOC 2, HIPAA, GDPR or ISO 27001 certification. Its
  compliance tooling helps you build your own evidence.
- Matimo is an MCP client (it connects to external MCP servers). It does not
  expose its own tools as an MCP server.
- Matimo Governance and Matimo Gateway are available on Matimo Enterprise.
- Not a fit if you need a fully autonomous agent with zero human checkpoints.

## Pricing

Plans and prices come from the live catalog. For current pricing, check
[matimo.ai/#pricing](https://matimo.ai/#pricing). Free to start, no card
required. You bring your own LLM key; Matimo never marks up model costs.

## Get started

- [Documentation](/docs): overview of each product, with status
- [API reference](/api) and [OpenAPI spec](/openapi.json)
- [About](/about) and [Contact](/contact)
- [Sign up](https://matimo.ai/signup)

## For agents and crawlers

- [llms.txt](/llms.txt): concise navigational index
- [llms-full.txt](/llms-full.txt): this overview plus every curated docs page in one file
- [Docs index](/docs/llms.txt)
