Graph-based, agent-first AI governance platform that connects AI systems, models, agents and datasets to curated risks, controls and policies.
At a Glance
About Saidot
Saidot is an AI governance platform from Saidot Ltd., a Helsinki-based company. It connects an organisation's AI inventory to a knowledge graph of risks, policies, models and controls, so systems and agents inherit risks and controls as they change. Saidot says the library covers 260+ risks, 620+ controls, 110+ policies and 170+ third-party AI models and products.
What It Is
Saidot is a web-based AI governance and compliance platform. Teams register AI systems, agents, models and datasets in a central inventory. The platform recommends relevant risks, controls and policy requirements based on classifications and linked components. It is aimed at AI teams, legal, compliance, risk and sourcing teams, and business leadership.
How the Graph Works
Governance is organised around connected entities: systems, models, agents, datasets, risks, controls and policies. A system uses models, agents and datasets. Risks are mitigated by controls, and controls map to policy requirements. Governance work done once at the model, agent or dataset level surfaces on every system that uses it, which drives risk and control inheritance and evidence reuse.
Agent Governance and Agent-First Workflows
An Agent Catalogue records each agent's purpose, lineage and autonomy level, and classifies the risk of the tools an agent can access. Users can connect their own AI assistants and agents through Saidot's MCP servers to draft risk assessments, register systems and research the library. Humans stay in control: agents surface and draft, and people decide and approve.
Compliance, Automation and Monitoring
Curated controls and guided workflows support the EU AI Act, ISO/IEC 42001, NIST RMF and 110+ other policies. Automation templates cover EU AI Act classification, risk inheritance and approval routing. Runtime events can trigger incidents, reviews, webhooks, and Slack or Jira alerts. The platform also exports audit-ready transparency reports and connects evaluation results to risk decisions. It connects to Azure AI Foundry and Amazon Bedrock and to other platforms via REST APIs, and Saidot says 95% of its functionality can be operated through APIs. Saidot states its ISMS is ISO/IEC 27001:2022 certified.
Community Discussions
Be the first to start a conversation about Saidot
Share your experience with Saidot, ask questions, or help others learn from your insights.
Pricing
Saidot Library
Holistic Library of AI knowledge
- AI policies, standards and laws
- AI risks and controls
- AI model cards and risks
- AI product cards and risks
Saidot Platform - Enterprise
Saidot Platform for up to 30 AI systems
- Saidot Library - library of AI knowledge
- Inventory of up to 30 AI systems
- Risk and compliance management
- Workflow automations
- Agent, model and dataset catalogues
- Saidot API and MCP servers
- More volume tiers for inventory scaling
Capabilities
Key Features
- Centralised AI inventory
- Knowledge graph of risks, controls, policies and models
- Automatic risk and control inheritance
- Agent Catalogue with tool-level risk classification
- Agent-first governance via MCP servers
- AI risk management
- Control and policy catalogue (EU AI Act, ISO/IEC 42001, NIST RMF)
- Workflow automations
- Runtime monitoring and event-driven governance
- Audit-ready reviews
- AI transparency reports
- AI evaluations
- REST API
Integrations
Demo Video

