One platform to govern
and operate your AI agents.

Matimo is ROAIQ's AI Agent Governance and Operations Platform. It rests on three pillars: Govern with Gateway, Governance and Guardrails; Operate with Workbench and Studio; and an Open Source foundation in Matimo OSS and AGDK that brings agents built anywhere into the same control plane. Start with the module you need and add the rest without rebuilding anything.

The Matimo platform in 30 seconds

Govern. Operate. Open Source.

Three pillars, one agent identity, one audit trail. Every module shares the same policy and identity model.

Govern

Control what agents say, call and do.

Gateway
One governed endpoint for every LLM call: identity, policy checks, routing, audit. Your keys.

Governance
Policy engine, Shadow Mode, Emergency Stop, agent identity, SSO/SCIM, compliance evidence exports.

Guardrails
Content-safety screening for conversations and MCP tool calls. Part of Workbench.

Operate

Build, run and monitor your agents.

Workbench
Multi-tenant agent execution: 16 reasoning engines, knowledge bases, HITL gates, cost tracking.

Studio
Visual workflow builder, included in every Workbench plan.

Open Source

The MIT-licensed foundation. Free, usable on its own.

OSS
The tool-execution and governance SDK the platform is built on.

AGDK
Identity, telemetry and per-tool-call checks for LangChain, CrewAI, Google ADK, AutoGen and plain Python agents.

Platform module · Live
Matimo Gateway - LLM Proxy and Router

One endpoint for
every governed LLM call.

Matimo Gateway is the LLM proxy and router for governed agents, and the platform AGDK checks in with. Change the base URL in an OpenAI-compatible or Anthropic-compatible client and every call is identified, policy-checked, routed, and audited, using your own provider keys.

Available to tenants with an active Matimo Enterprise license.

Govern module · Enterprise add-on

Matimo Governance

The module that wraps around your existing Workbench agents. Zero code changes. Activate in Shadow Mode first: see exactly what governance would block before you block anything. Flip to Enforce when you're ready.

Shadow Mode
All policies evaluate. All decisions are logged. Nothing is blocked. You see the blast radius of every rule before it goes live. Then flip a single switch to enforce.
  • 7-step execution gate on every agent action: license, Emergency Stop, identity, lifecycle, budget, spawn depth, policy rules
  • Emergency Stop: one-click kill switch for all agents, instantly
  • Agent Soul documents: versioned identity and behavioral contract per agent
  • Policy engine: 16 built-in templates + custom YAML rules + simulation
  • Governance Health Score: 0–100 grade for your entire agent fleet
  • SAML 2.0 SSO with JIT provisioning (OIDC, LDAP and SCIM built, live verification pending)
  • Content inspection: PII detection, secret scanning, prompt injection scoring
  • Compliance evidence reports for six frameworks (SOC2, HIPAA, GDPR, EU AI Act, ISO 27001, NIST AI RMF)
  • Webhook / SIEM streaming: Splunk, Datadog, or any HTTP endpoint
Request Matimo Governance →
Matimo Workbench Operate module · Generally available
Matimo™ Workbench: Agent Execution Platform

Operate AI agents at enterprise scale

Matimo Workbench is ROAIQ's production platform for governed AI agent deployment - multi-tenant, fully auditable, and built around the governance requirements enterprise operations teams actually enforce.

  • 16 reasoning engines: ReAct, Chain-of-Thought, Tree-of-Thoughts, and 13 more
  • 5 LLM providers: OpenAI, Anthropic, Google, Ollama, AWS Bedrock, no lock-in
  • 129 ready-to-use tools via Matimo OSS (Gmail, Slack, GitHub, HubSpot and more), plus a governed 449-tool Composio catalog
  • Session + long-term memory with semantic vector search: agents that remember
  • Knowledge bases with RAG: agents grounded in your own business documents
  • HITL safety gates: humans approve before any consequential action runs
  • Matimo Studio (19 node types) + REST API + webhooks + scheduled execution
  • Semantic cache: reuses LLM responses by meaning to reduce repeat LLM calls
Matimo Workbench / Studio - Procurement Agent EXAMPLE
⚡
API Trigger
Tenant: globex-corp · POST /v1/agent/procure
DONE
◈
Intent Classifier - Claude 3.5
Engine: reasoning-v3 · 417 tokens
RUNNING
⚙
HubSpot - Update Deal Stage
Governed by: policy/crm-write-approval
QUEUED
↗
Structured Output + Slack Notify
Audit: logged · Schema: ProcurementResult v1
QUEUED
16
Reasoning Engines
129
Tools Built-in
5
LLM Providers
matimo.yaml
tools:
  - name: send_email
    description: "Send via Gmail"
    provider: gmail
    params:
      to: string
      subject: string
      body: string

  - name: create_issue
    description: "Open GitHub issue"
    provider: github
    params:
      repo: string
      title: string

policy:
  tool_access:
    role/analyst:
      - send_email
    role/admin:
      - "*"
MIT
License
129
Ready-to-use Tools
10
Provider Packages
Matimo OSS Open Source · MIT
Matimo OSS - Open-source foundation

Define tools once.
Govern everywhere.

Matimo OSS is ROAIQ's open-source agent governance library. Define your integrations once in YAML and use them across LangChain, OpenAI, Claude, CrewAI, and any other AI framework - with built-in policy enforcement, tool RBAC, and audit logging.

  • Framework-agnostic - write once, use everywhere
  • YAML-first schema with TypeScript type generation
  • 10 built-in providers: Gmail, GitHub, Slack, Notion, Microsoft 365…
  • Policy registry: tool RBAC, agent identity, audit log
  • npm + PyPI - install in seconds, no infrastructure
  • v0.2.0: governance on by default, tamper-evident audit log, same rules in TypeScript and Python
  • MIT licensed - free forever, no vendor lock-in
Matimo AGDK Open Source · MIT · v0.1.0
Matimo AGDK - Agent Governance Development Kit

Govern the agents
you already built.

Matimo AGDK is an open-source Python SDK that brings agents built in LangChain, Google ADK, CrewAI, AutoGen, or plain Python under the same governance as native Matimo Workbench agents. It never holds your LLM provider key and never executes your tools.

  • Cryptographic agent identity, signed locally
  • Mandatory telemetry: an agent that goes quiet is provably absent
  • Per-tool-call policy check, with human approval when a policy requires it
  • Python 3.13+, MIT licensed, available on PyPI

Which module do you need?

Every module of the Matimo platform solves a different problem. Matimo Workbench runs your agents. Matimo OSS powers its tools. Matimo Governance adds enforcement and evidence. Most regulated enterprises use all three.

Capability Matimo OSS Matimo Governance Matimo Workbench
License / Cost MIT · Free Paid · Contact us Freemium + Pro + Enterprise
Agent building + deployment – – ✓ Matimo Studio + API
LLM providers – – 5 providers, no lock-in
Ready-to-use tools 129 tools, 10 packages (+449 via Composio) ✓ Included ✓ Built-in
Knowledge bases + RAG – – ✓
HITL safety gates – ✓ ✓ Built-in
Agent governance + RBAC ✓ Tool-level ✓ Full execution gate ✓ Native
Shadow Mode (observe before enforce) – ✓ Enterprise only –
Emergency Stop (all-agent kill switch) – ✓ Enterprise only –
Agent Soul + identity tokens – ✓ Enterprise only –
Full execution audit trail Tool-level ✓ Every gate decision ✓ Every token + tool call
Compliance evidence (6 frameworks) – ✓ Enterprise only –
SAML 2.0 SSO / SCIM – ✓ Enterprise only Enterprise tier
Execution cost tracking – – ✓
Open source ✓ MIT – –
Status ✓ Live on npm ✓ Generally available ✓ Generally available

Not sure where to start?

Our engineers will assess your architecture in 30 minutes and give you an honest recommendation - even if the answer is "just use Matimo OSS for now."