Saidot is an AI governance platform that helps organisations keep track of the AI systems and agents they build or buy, understand the risks each one carries, and show regulators and auditors that the right controls are in place.
It is built for governance, risk and compliance teams working towards the EU AI Act, ISO/IEC 42001 and the NIST AI RMF. Rather than starting every assessment from a blank form, teams link each AI system to the models, datasets, tools and agents it uses, and the platform brings in the relevant risks, policies and controls from a curated library.
At A Glance
| Vendor | Saidot Ltd. |
|---|---|
| Category | AI Governance & Compliance |
| Identities covered | AI systems, models, datasets and AI agents, as governance records |
| Deployment | Web application, also available through Microsoft Azure Marketplace |
| Agentic AI | Agent catalogue, tool risk classification, three MCP servers |
| Customers | City of Amsterdam, City of Helsinki |
| Certifications | ISO/IEC 27001:2022 |
| Founded | 2018, Helsinki |
| Market classification | AI Governance & Compliance |
What Saidot Does
Most AI systems are assembled from parts: a third-party model, a dataset, a few tools and, increasingly, an agent that acts on its own. Saidot governs the whole assembly. Each system in the inventory carries links to its components, and risks and controls are inherited from those components, so when a model or tool changes the governance record changes with it. Every classification, control and approval is recorded so it can be traced during an audit. The company’s guiding belief is that AI can be “a force for good if used responsibly”.
Products
Saidot Graph
The Graph is the knowledge base behind the platform. It links customers’ AI systems, models and agents to more than 260 risks, 620 controls and 110 policies curated by Saidot’s experts, and it profiles more than 170 third-party AI models and products. When a system is linked to a model or dataset, the risks and controls attached to that component flow through automatically.
AI Inventory
The inventory gives a governance team one place for every AI system, model, agent and dataset in the organisation, and it updates itself as components change. Agents can be imported from Microsoft Foundry Agent Service and Amazon Bedrock, and models from Azure OpenAI Service, Azure AI Services and Azure ML, which saves teams from registering cloud AI assets by hand.
Risk Management And Compliance
This module turns the inventory into evidence. Each risk is scored for likelihood and impact on a scale of 1 to 5, controls are managed centrally with risk mapping and default evidence specifications, and controls and evidence can be reused across systems instead of being rebuilt each time. Transparency reports for auditors or the public are generated directly from the system record.
AI Evaluations And Runtime Monitoring
Saidot generates testing and red teaming plans from a system’s context, tasks and the risks of its linked models, so evaluation effort goes where the risk sits. Once a system is live, runtime observability events from the customer’s AI stack are ingested and turned into governance actions.
How It Works
Saidot runs as a web application at app.saidot.ai and can also be bought through the Microsoft Azure Marketplace. Governance data is held as a knowledge graph that is open to people, software and agents alike. Automation handles risk inheritance, control inheritance and lifecycle approvals, and around 95% of the platform can be operated through the REST API, so governance steps can be built into existing delivery pipelines. Webhooks push changes out to other tools, and three MCP servers give AI agents direct access to the same data.
Identities And Systems Covered
| AI systems | Registered in a central inventory with lineage to models, datasets and tools. |
|---|---|
| AI agents | Recorded with purpose, lineage and autonomy level, with a risk classification for each tool they can reach. |
| Models | The library’s third-party model profiles plus models imported from Microsoft’s Azure AI services. |
| Cloud AI platforms | Azure AI Foundry, Microsoft Foundry Agent Service and Amazon Bedrock. |
Agentic AI Capabilities
Agent Governance
Agents are governed below the model level. Customers can set a risk classification for each tool an agent can access, which reflects the fact that an agent’s risk depends on what it can do as much as on the model behind it.
Agent-First Governance Through MCP
The Docs, Library and Governance MCP servers let a customer’s own assistants and agents research the library, draft risk assessments, register systems and pre-fill metadata. People stay in charge of decisions: “agents surface and draft; your team reviews, decides, and approves where it matters”.
Integrations
| Area | Integrations |
|---|---|
| Collaboration and ticketing | Slack, Jira, Atlassian |
| Data | Snowflake |
Company Background
| Founded | 2018. |
|---|---|
| Headquarters | Helsinki, Finland, with a remote-first team. |
| Funding | A €1.75 million seed round in October 2023 led by Crowberry Capital with Ventic. |
| Certification | ISO/IEC 27001:2022 certified information security management system, audited by Prescient Security. |
| Partnership | A collaboration with Microsoft links Azure model registries to the Saidot Model Catalogue. |
Note: This overview is compiled from Saidot’s publicly available information as of 3 November 2026. It is provided for information only. NHI Mgmt Group does not endorse any vendor, and nothing on this page represents the opinions of NHI Mgmt Group.