Deeploy
Utrecht-based AI governance platform that connects to existing MLOps and GenAI stacks and enforces controls on systems and agents at runtime
Quick Overview
| Company | Deeploy |
|---|---|
| Category | AI Governance and EU AI Act Compliance Software |
| Headquarters | Utrecht, Netherlands |
| Founded | Not stated |
| EU Presence | Yes - Netherlands (EU) |
| Data Location | SaaS, private cloud, own cloud (AWS or Azure) or on-premise on Enterprise |
| Open Source | No |
| GDPR Compliant | Yes |
| Self-Hosting | Enterprise plan: own cloud (AWS or Azure) or on-premise |
| Main Features | Control frameworks (ISO 42001, NIST AI RMF, custom), risk classification, assessments with auto-collected evidence, AI discovery, drift alerts, tracing, guardrails, agent registry and control |
| Pricing | Platform fee plus seat fee by stage, no amounts published; Core, Scale and Enterprise plans |
| Best For | Teams that already run models and agents and want governance enforced where they run |
| Replaces | Credo AI, manual AI registers |
Detailed Review
Deeploy sits closer to the running systems than most governance tools. It connects to MLOps and GenAI platforms, finds AI that was never onboarded, and then monitors the systems for drift, performance drops and unsafe outputs.
The privacy policy names Deeploy, established at Oudegracht 91A in Utrecht, Chamber of Commerce number 78193621, as controller.
Controls Tied to Frameworks
Default frameworks include ISO 42001 and the NIST AI RMF, and custom frameworks are possible. Each AI system gets a risk classification and an approval process, and assessments collect evidence automatically.
The site also runs an EU AI Act hub for teams preparing for the regulation.
Agents as Governed Objects
An agent governance area registers, traces and controls AI agents across the organisation. A Model Context Protocol server was announced for agentic systems.
The depth of the agent controls could not be tested from public pages, so a demo should check which actions can be blocked and which only logged.
Where It Runs
The plans table shows Core as SaaS only, Scale as SaaS or private cloud, and Enterprise with a custom set-up. The FAQ adds that Enterprise can be deployed inside your own cloud (AWS or Azure) or on-premise, which suits teams with strict data rules.
Limitations to Consider
No amounts appear on the pricing page: it is a platform fee plus a seat fee by company stage, with a minimum of 3 seats on Core and 5 on Scale. The Core plan has no SSO and no SLA. The focus on models and runtime means business-side workflows such as legal intake are lighter than in Enzai. The pages checked do not say where the SaaS version is hosted.
Who Should Use Deeploy
Dutch and European AI and risk teams in regulated sectors that need governance connected to the MLOps stack they already use.
Pros and Cons
Pros
- Connects to existing MLOps and GenAI platforms
- Own-cloud and on-premise option on Enterprise
- Drift monitoring and guardrails included
- Agent registry and tracing
- Clear plan structure with stated minimum seats
Cons
- No published amounts
- Core plan has no SSO or SLA
- SaaS hosting region not stated
- Less focus on business-side intake workflows
Alternatives to Deeploy
Other European ai governance and eu ai act compliance software worth comparing before you decide:
Saidot
Finland · Governance platform on a knowledge graph of AI risks and controls
Modulos
Switzerland · Zurich platform that prices AI risk in euros and tracks controls
Validaitor
Germany · Karlsruhe platform joining AI inventory, red-teaming and compliance
Enzai
United Kingdom · Shadow AI discovery, intake review and agent action control
See all European ai governance and eu ai act compliance software or compare them directly against Credo AI on our alternatives page.
Frequently Asked Questions
On the Enterprise plan, yes: the pricing FAQ says it can be deployed inside your own cloud (AWS or Azure) or on-premise.
A monthly platform fee plus a seat-based fee, both depending on company stage, and a custom Enterprise plan. No amounts are published.
Yes. The product has an agent governance area that registers, traces and controls agents, and the vendor announced an MCP server for agentic systems.