Collibra Launches Runtime Governance Tools to Constrain Enterprise AI Agents

The shift toward autonomous agents that take action across enterprise infrastructure has made runtime governance essential. Companies must now manage both the information context that agents access and the operational limits within which they function.

Collibra BV has unveiled four new capabilities—Live Map, Maestro, Guardian Agents and Agent Contracts—to address this governance challenge. The company frames the problem as the "hallucination tax," referring to costs incurred through manual verification, rework and risk mitigation. Felix Van de Maele, chief executive officer and co-founder of Collibra, explained the fundamental shift: "I think the stakes just have gotten so much higher. The big difference today versus two or three years ago is that we used to have people in the middle, and we could count on the judgment of people. That has gone away, and so governance has moved from documentation, design time [and] policy setting to runtime."

Live Map builds reusable context for agent fleets

Graph databases are emerging as a critical context layer for enterprise agents, though Van de Maele cautions against treating them as a one-size-fits-all retrieval solution. Depending on the specific task, agents may draw from vector stores, structured files, semantic search capabilities or graphs grounded in business ontology.

"There's not going to be one mechanism that is able to answer every potential question," Van de Maele said. "The question is, 'How do we route the right context to the right retrieval mechanism at the right time?' Graph is key."

Live Map extracts reusable context from curated unstructured documents, eliminating the need for agents to reconstruct relationships with each query. Collibra's Knowledge Graph, built on Neo4j, enables traversal of relationships across enterprise assets.

"You have a lot of unstructured documents from your SharePoint. What you don't want is for the model to [parse through the data] for every single question," Van de Maele explained. "Building, engineering that context and making that available through Live Map, you hit it once offline, and so you have the answer ready every time. That's a massive token efficiency."

Runtime governance turns policies into enforcement

Collibra Maestro applies agents to governance functions that organizations have traditionally managed through manual stewardship, designed to help governance teams keep pace with expanding artificial intelligence deployments.

"The challenge is, how do you govern at the speed of AI, which is different than the way we had to govern five years ago?" Van de Maele said. "Maestro gives our customers the ability to build these governance steward agents that automate up to 80% of the governance work. You're able to govern quickly and efficiently enough to keep up with the demands of new AI workloads."

Agent Contracts specify permissible agent behavior by considering data sensitivity, operating context and risk factors. Guardian Agents then enforce these contracts before any action is executed.

https://www.youtube.com/embed/FdtCk9IbVxQ?feature=oembed

"We've all read the examples where the agent escapes," Van de Maele said. "The way we're solving that is by encoding and compiling these Agent Contracts and deploying those into your existing AI gateways or AI orchestration tooling. Before the agent is able to take the action, we monitor: Does it align to the Agent Contract? If not, we block it, and the agent never has the opportunity to take that action."