In its first major announcement since being acquired by Salesforce Inc., data management vendor Informatica today unveiled a “headless” version of its flagship Intelligent Data Management Cloud, positioning it as a resource for enterprises building artificial intelligence agents that need governed, context-rich data without relying on traditional application interfaces.

“Headless” is a software architecture in which the user interface is separated from the back-end data and logic. The processing engine then operates independently, communicating with other applications via application programming interfaces.

Informatica said it will expose IDMC through reusable services and Model Context Protocol endpoints that can be invoked directly by AI agents, developer tools and enterprise workflows. The company said the approach allows customers to access governance, integration, data quality and master data management functions from environments that initially include Anthropic PBC’s Claude, Salesforce’s Slack, Anysphere Inc.’s Cursor and other AI frameworks without extensive integration.

The company also introduced what it described as the industry’s first unified “Agent and Context Catalog,” intended to govern both enterprise data assets and AI agents from a single control plane. Informatica executives framed the moves as a response to growing concerns that enterprise AI initiatives are being constrained by fragmented and poorly governed data.

“You can’t have trusted AI without trusted data,” the company said in the release, citing its own chief data officer survey which found that 76% of data leaders said governance has not kept pace with AI initiatives and 61% said higher-quality data makes it easier to move AI pilots into production.

Easier discovery

The headless architecture is less about replacing APIs than making enterprise data services easier for agents to discover, interpret and orchestrate, said Rahul Auradkar, president of data foundations at Salesforce.

“With the power of AI and the standardization that you have with command-line tooling, MCP and APIs, it’s significantly easier to use the power of AI,” he said. “The building blocks now became more standardized, easier to access and smarter to combine and process.”

Gaurav Pathak, Informatica’s senior vice president for metadata and AI products, said the goal is to give agents substantially more contextual awareness than traditional API integrations provide.

“We want to give agents a whole lot more context,” he said. “What is the best way to invoke these APIs? What are the right arguments to pass? What are the edge conditions?”

A major component of the strategy is Claire, Informatica’s AI-powered data management engine that uses machine learning and metadata intelligence to automate repetitive data tasks. Claire will now operate as a “fully headless, multi-agent intelligence layer,” the company said. It also announced several new autonomous agents, including a Data Quality Agent that generates production-ready quality rules from natural language prompts and a Metadata Enrichment Agent that automatically fills gaps in enterprise catalogs.

Another addition is “Agentic Multidomain MDM,” which continuously cleanses, stewards and enriches master data for use by autonomous agents in real time.

Pathak said the company has seen dramatic productivity gains during preview deployments of the Data Quality Agent.

“Earlier, organizations would go at max about four or five data quality rules a week,” he said. “We’ve seen organizations go up to 200 data quality rules a day with the agent.”

Executives emphasized that Informatica’s metadata catalog, which is considered a core competitive asset, is central to its AI strategy. Metadata is a “table of contents” for enterprise information that helps agents determine which data is most relevant and trustworthy.

“One of the core unique selling points for Informatica is our unified metadata foundation,” Pathak said. “No agent can take all of this encyclopedia in their context window at one time. The table of contents guides them.”

Safeguards

The company also said it has paid heed to balancing automation with governance safeguards. Agents will use deterministic workflows only when outcomes are predictable and seek human oversight when ambiguity exists.

“Pretty much everything that we do that is nondeterministic has a human-in-the-loop override if there’s a need for it,” Auradkar said. “The key is reliable outcomes.”

Informatica’s broader strategy increasingly aligns with Salesforce’s push toward so-called “agentic enterprises,” in which AI agents interact across business systems, collaboration platforms and analytics tools using shared governance and data controls. The new Agent Fabric Context Catalog integrates Informatica governance capabilities with Salesforce’s MuleSoft Agent Fabric to govern both agents and data assets together.

The company is also announcing a series of partnership expansions with major cloud and data platform providers that tie Informatica’s technology closer to their offerings.

Claire GPT will be available in Google Cloud environments with added support for Google’s emerging Agent2Agent interoperability protocol. That allows Claire agents to collaborate with agents built on Google Gemini Enterprise, Informatica said.

An expanded partnership with Snowflake Inc. includes headless data management integration with Snowflake Cortex AI, row-level governance for Snowflake tables and metadata scanning for managed Apache Iceberg tables.

Microsoft Corp. will support Informatica’s MCP servers in the Microsoft Foundry platform for building, testing, deploying and managing AI applications and agents. The firms also pledged to deepen integration between IDMC and Microsoft Fabric for large-scale ingestion and change-data-capture workflows.

Expanded integrations with Amazon Web Services Inc. will expose Claire agent skills and MCP servers through the AWS Agent Registry and Amazon Quick to support governed AI workflows on Amazon Bedrock.

Finally, an agreement with Databricks Inc. features new integrations with that company’s Agent Bricks, connections to the Lakebase database management systems, master data publishing into the Databricks lakehouse and governance federation with Unity Catalog.

Image: SiliconANGLE/Google Flow