Salesforce extends its headless push into enterprise data via Informatica

Salesforce appears to be on a mission to “decapitate” enterprise software. After unveiling headless commerce and headless applications, the company late on Wednesday extended the label to enterprise data management via Informatica, one of its more recent acquisitions, as part of its broader industry push to prepare enterprise systems for autonomous AI agents.

While Salesforce has spent the past year stitching together Agentforce, Data Cloud, MuleSoft, Tableau, and its Customer 360 portfolio into what it describes as a unified AI operating layer, the headless architecture from Informatica, it said, fills a major gap around trusted enterprise data, including governance, metadata management, and lineage, all of which are required for AI agents to act with less human supervision.

As part of this new architecture, Informatica said that it is effectively breaking apart its traditional Intelligent Data Management Cloud (IDMC) platform into reusable data management services that can be invoked directly inside AI-native environments rather than primarily through Informatica’s own interface.

“Enterprises can invoke data management operations directly from their favorite LLM or IDE,  including Claude, Slackbot, Cursor, and more, bringing trusted data management into the tools and workflows where developers and agents already work,” the company said in a statement.

According to Bradley Shimmin, lead of data and analytics practice at The Futurum Group, that invocation from inside IDEs and vibe coding tools is made possible because those reusable data management services are essentially being exposed through Model Context Protocol (MCP) rather using than custom connectors.

Reduced learning curve for developers

For developers, IDMC’s shift from being a separate visual platform to an invisible utility layer quietly running in the background should enable faster buildout of automated workflows, as they will be able to skip the tedious process of writing custom integrations, Shimmin said.

“And,” he added, “because it relies on standard communication protocols, embedding these capabilities into custom applications requires significantly less effort than traditional methods.”

Ashish Chaturvedi, leader of executive research at HFS research, also pointed out that the same shift would also meaningfully reduce the learning curve for developers, from weeks to minutes.

Implications for CIOs

Additionally, said Mike Leone, principal analyst at Moor Insights and Strategy, the architectural change carries broader implications for CIOs.

“By embedding governance, lineage and policy controls directly into reusable services rather than confining them to Informatica’s standalone console, enterprises can ensure those guardrails follow every agent and workflow that developers create, which is exactly what CIOs need as agent sprawl accelerates,” he said.

However, Chaturvedi warned that the same architecture enabling enterprises to more easily deploy and orchestrate large numbers of AI agents could also sharply increase infrastructure and consumption costs, making it important for CIOs to carefully evaluate pricing and operational models before scaling agentic AI initiatives.

Claire’s extension outside IDMC to simplify DevOps

Claire, IDMC’s metadata-based AI engine, is also being repositioned under the broader headless strategy, Informatica said, adding that it can now be accessed from other platforms, such as Salesforce, AWS, Microsoft, Databricks, and Snowflake.

For developers, said Leone, the ability to invoke Claire outside of Informatica’s UI means that they can perform natural language-driven data operations directly inside the environments where they are already building applications and workflows.

For CIOs, however, the governance posture remains consistent, said Robert Kramer, managing partner at KramerERP, because Claire continues operating against the same underlying metadata foundation regardless of where agents or developers invoke it.

Both the new headless architecture and the updated Claire are expected to be generally available by the end of Q2.

Reducing governance complexity

Alongside the new headless architecture, Informatica also introduced Agent Fabric Context Catalog, which it described as a centralized governance and discovery layer for data assets and AI agents that expands on its broader push to build machine-readable context layers for autonomous AI systems.

“For the first time, enterprises can see, govern, and trust everything running in their AI ecosystem: every vetted AI agent, every enterprise data asset, and the governance policies that control how agents are built, deployed, and act, with full lineage at the point of integration,” the company said.

According to Kramer, the product appears to align closely with Salesforce’s broader MuleSoft Agent Fabric strategy, which provides centralized discovery, governance, and orchestration for AI agents and MCP services. Informatica effectively becomes the trusted enterprise data, metadata, lineage, and governance layer underneath those agent workflows.

For Amit Chandak, chief analytics officer at AI, data engineering, and migration consultancy firm Kanerika, the unification of two separate catalogs, one for data and one for agents, will help enterprises reduce governance complexity, especially in the wake of agent proliferation and possibility of agent sprawl.

Agent Fabric Context Catalog is expected to become available by the end of Q2, according to the company.

Informatica also announced what it described as industry’s first agentic multidomain master data management (MDM) system that it said has been designed to let AI agents continuously cleanse, steward, and enrich master data in real time instead of relying on slower manual MDM processes.

According to Constellation Research Principal Analyst Michael Ni, this system, too, aligns with Salesforce’s broader autonomous workforce strategy, as AI agents will require an always-on trusted operational data layer.

For developers and enterprise teams, the feature can provide a “massive lift”, Moor Insights and Strategy’s Leone pointed out; he feels that MDM in particular has traditionally been among the most manual and steward-intensive parts of enterprise data operations.

However, he noted, developers will still need to define  governance and stewardship rules in plain English to ensure the agents behave as intended.

The catch, the analyst added, will be accuracy: “If the agents are off, you’re injecting more risk with less visibility.”

The new MDM system, along with a new Data Steward Agent, are expected to made available by the end of the year.