Collibra AI Agent Registry now in preview.
***
As AI adoption accelerates, organizations face mounting challenges in managing compliance, traceability and lifecycle efficiency. According to Gartner’s 2023 AI in the Enterprise Survey, it takes an average of 8.2 months to move an AI prototype into production. And only 7% of companies can do it in under three months.
AI model governance: Introducing new capabilities for greater control and accountability
As AI development accelerates, many organizations struggle with visibility and accountability. Models are created across teams, tools and environments, with inconsistent documentation and unclear ownership. That makes it difficult to meet regulatory requirements, align with business objectives or ensure responsible outcomes.
To address this challenge, the Collibra AI Model Registry provides customers with a unified platform to manage all AI models while linking each one to its underlying data, policy context and business stakeholders. Additionally, our expanded integration across the AI ecosystem—with new support for Azure AI Foundry and MLflow—builds on existing connections to AWS Bedrock, AWS SageMaker, Databricks Unity Catalog, Google Vertex AI, Microsoft Azure ML and SAP AI Core.
The result: Data and AI professionals get transparency from day one and support for proactive risk and compliance management throughout the model lifecycle.
In parallel, we’ve introduced lifecycle sign-offs, allowing teams to embed formal checkpoints across the development process. These structured reviews capture ownership, rationale and evaluation outcomes, ensuring models are reviewed and approved before moving forward. And the benefit is twofold: A more predictable path from experimentation to production, without compromising on trust.
With the introduction of the AI Model Registry, we’re strengthening the power of AI governance already embedded in the Collibra platform.
Looking ahead: AI agent oversight in preview
AI agents are increasingly used to automate service interactions, business processes and content creation. But unlike traditional models, agents are autonomous and act autonomously on behalf of the enterprise, often making decisions without human intervention. Whether developed in-house or purchased from third parties, the use of agentic AI introduces a new level of risk, demanding new forms of governance and oversight.
Get a preview of the Collibra AI Agent Registry
To support the rise of autonomous systems, the Collibra AI Agent Registry is now available in preview. A tool-agnostic capability, the AI Agent Registry enables organizations to associate each AI agent with its specific business context, designated owner, lifecycle stage and policy alignment, eliminating blind spots and laying the groundwork for a reliable, traceable and compliant use of AI agents within enterprise environments.
The preview launches with an initial integration with Azure AI Foundry—our first supported environment for agent oversight—and additional integrations will be added over time based on customer feedback to expand coverage and ensure alignment with enterprise needs.
