Collibra, the data governance company that now describes itself as an enterprise AI control plane, announced on October 5, 2026 that it had acquired trail ML, an AI governance startup based in Munich.
Neither company disclosed the price or other terms. Collibra's announcement did not say whether trail ML's product will keep its own name or how many of its staff are joining, but trail ML's website now carries a banner saying the company is part of Collibra.
The deal gives Collibra software that checks AI systems against regulatory and industry frameworks on an ongoing basis, rather than through one-off reviews before a model or agent goes live.
What trail ML built
trail ML started out as an AI governance copilot: a platform that gathered and organised the documentation around AI projects so that engineers, lawyers and risk teams could work from the same information. Early on it pitched itself at companies preparing for the EU AI Act and for ISO 42001 certification, and it offered a free self-assessment tool that sorts an AI use case into its EU AI Act risk class.
By the time of the acquisition the product had moved toward agents. According to Collibra, trail ML's agents read the context and evidence around an AI system, work out which frameworks and controls apply, assess whether those controls exist and work, flag gaps, and run the follow-up governance workflows. Solutions Review reported that the software also triggers a fresh assessment when the evidence behind a system changes.
The startup also built runtime enforcement. Collibra said these capabilities apply its policies at the point where AI agents run and stop actions that would violate them before they happen. The frameworks named in the announcement include the EU AI Act, ISO 42001 and the NIST AI Risk Management Framework.
Unite.AI, summarising trail ML's own materials, reported that the software connects to tools such as Confluence, Jira, GitHub, Databricks, MLflow, Hugging Face, ServiceNow and OneTrust, and can run as EU-hosted SaaS, on premises or in a customer's own cloud. trail ML has claimed 70% faster compliance work for customers; those figures come from the company and have not been independently checked.
“AI governance cannot become another manual process that slows organizations down as AI scales. trail ML brings powerful automation to Collibra's context and control.”
Why Collibra is buying
Collibra built its business on cataloguing and governing enterprise data, and it has spent the past two years extending that pitch to AI models and agents. Its argument in the announcement is that knowing which policies apply is no longer enough: companies need to keep checking that controls work as AI tools touch their data and systems.
Co-founder and chief executive Felix Van de Maele said manual governance would not keep pace with AI adoption, and that trail ML would mean less manual work for governance, risk and compliance teams in assessing controls and keeping evidence up to date. CDO Magazine described the deal as a move from reviewing AI before deployment toward governing it while it runs.
The purchase follows Collibra's acquisition of Raito, a data access governance company, which it announced in June 2025 as part of a push to control access for both people and AI agents.
The founders and the funding
trail ML was founded in 2023 by Anna Spitznagel, who is chief executive, together with Nikolaus Pinger and Sven Hölzel. The three met while studying at the Technical University of Munich and the Center for Digital Technology and Management (CDTM).
In July 2024 the company raised a €1.45 million (about $1.7 million) pre-seed round led by the early-stage investor CapitalT. NP-Hard Ventures, Mozilla Ventures, Common Magic and several business angels also took part. No later round has been reported.
Spitznagel said the founders started trail ML because they believed AI governance had to be automated and operational, and that combining it with Collibra's enterprise context and reach would let organisations move faster with AI.
A busy market for AI governance
Collibra itself was founded in Belgium in 2008 and was valued at $5.25 billion after a $250 million Series G in November 2021, led by Sequoia Capital Global Equities and Sofina, according to the company. It has not announced a newer valuation.
Demand for AI governance tooling has grown as the EU AI Act's obligations phase in and as companies give software agents access to internal systems.
