Dateline: October 6, 2026 — Data and AI governance company Collibra announced on October 5 that it has acquired Trail ML, a Munich-based AI governance startup founded in 2023, to add agent-powered automation and runtime control to its platform.
The short version: Collibra wants to automate the governance of AI systems the same way companies are automating everything else — with agents that assess controls continuously and block violations before they happen, instead of manual audits that go stale the day after they’re done.
What happened
Collibra, which describes itself as the enterprise AI control plane, announced the acquisition of Trail ML on October 5, 2026. Trail ML was founded in 2023 by Anna Spitznagel, Nikolaus Pinger, and Sven Hölzel. The companies did not disclose a price.
The deal brings Trail ML’s agent-powered automation into Collibra’s governance stack, covering what Collibra calls the two gaps in modern AI governance: continuous assessment of controls and runtime enforcement where agents actually run.
The details
Trail ML’s agents can examine the context around an AI system, identify which frameworks and controls apply, and evaluate whether those controls are in place and working — a way to operationalize requirements across the EU AI Act, ISO 42001, and the NIST AI Risk Management Framework. Because assessments can be re-triggered when evidence changes, companies can replace manual, point-in-time assessments with a compliance posture that stays current.
On the runtime side, Collibra said AI incidents increasingly start with agents taking actions rather than models giving answers. Trail ML’s runtime capabilities enforce Collibra policies where agents run and block actions that violate them before they happen.
Trail ML also works the other direction: it applies AI agents to governance, risk, and compliance work itself — automating risk categorization, gap analyses, control assessments, evidence collection, and report generation across a customer’s existing tools. Its agents only write to customer systems when a human approves, under a mechanism it calls “Copy-on-Write.” The company says its agents deliver 4x faster AI deployment and 70% faster compliance execution, figures it attributes to customer outcomes.
Why it matters
Governing AI with spreadsheets and annual reviews is losing viability by the month — agents that act at machine speed need governance at machine speed. The irony Collibra is banking on is the honest one: the only way to govern thousands of AI agents is to deploy AI agents to do the governing. Whether enterprise compliance teams will trust automation with their audit evidence is the real question here, but “Copy-on-Write” human approval suggests Trail ML at least understands the trust problem. For buyers drowning in EU AI Act obligations, an automated answer that actually integrates with Confluence, Jira, ServiceNow, and OneTrust instead of demanding a rip-and-replace has a real shot.
FAQ
What is Trail ML?
Trail ML is a Munich-based AI governance company founded in 2023 by Anna Spitznagel, Nikolaus Pinger, and Sven Hölzel. It builds agent-powered automation for AI governance and for compliance workflows themselves.
What does Collibra do?
Collibra describes itself as the enterprise AI control plane, providing the context and control organizations need to govern AI across data, models, applications, and agents.
What is “Copy-on-Write”?
It’s Trail ML’s approval mechanism: its agents only write changes to customer systems when a human approves them first.
Why acquire an AI governance company now?
Collibra argues it’s no longer enough to know which policies and regulations apply — organizations need to determine the relevant controls continuously, confirm they work, and enforce them as AI systems and agents interact with enterprise data and systems.
Sources: Unite.AI, Collibra