Informatica adds agents to automate its Intelligent Data Management Cloud

Informatica is preparing ready-built AI agents and agent-building tools to help enterprises automate their use of its Intelligent Data Management Cloud (IDMC).

Claire Agents will help enterprises use natural language to automate the management of data quality and governance, the company announced at its annual Informatica World conference in Las Vegas.

The agents can help discover data, understand data based on business semantics, create data pipelines to collect, integrate, and clean the datasets required, generate insights from datasets, generate a data model for master data, and generate ETL or  ELT pipelines, VP of product management Gaurav Pathak told Infoworld.

“Before the introduction of Claire Agents, each of the capabilities had to be carried out manually via code or through guided systems,” Pathak said. “Despite the use of guided systems earlier, enterprise users were required to set up business logic, perform software development lifecycle (SDLC) tasks, and maintain the workflows or applications,” Pathak added.

Data cleaning and augmentation are never-ending tasks for many enterprises, said Amalgam Insights chief analyst Hyoun Park, and automation could help them finally get on top of things.

The first of the new agents will tackle tasks including data quality, data discovery, data lineage, data ingestion, ELT, and data exploration, but Informatica customers will have to do things manually for a little longer: The company doesn’t expect to offer a preview of the new agents until the Fall.

Support for Anthropic’s MCP and Google’s A2A protocols

Informatica said that Claire Agents can act as an endpoint for Google’s recently released Agent2Agent (A2A) protocol, and as both client and server for Anthropic’s Model Context Protocol (MCP). Salesforce, Workday, SAP, MongoDB, and ServiceNow have adopted the A2A protocol, while MCP is an increasingly popular choice among other data management and analytics software providers.

As MCP clients, Claire Agents can use the MCP server implementations in third-party services such as catalogs to discover and explore metadata. As an MCP server, Claire exposes agent capabilities so that other MCP clients can access the agents and other IDMC capabilities, Pathak said. These capabilities include creating and managing data Integration pipelines, creating and enriching master data management (MDM) business entities, and monitoring IDMC workflows, he said.

There have been some concerns raised about challenges managing the security of MCP; Pathak said that Claire will use the security context of the user credentials provided to the API for MCP implementation, so it remains compliant with the security layers in the customer environment.

Building and orchestrating agents

AI Agent Engineering, Informatica’s service within IDMC for building, connecting, and managing intelligent multi-agent systems, will be available globally later this year.

The service provides a unified no-code environment for technical and non-technical users to build and orchestrate agents across ecosystems like AWS, Azure, Databricks, Google Cloud, Salesforce, and Snowflake among others, Sumeet Agarwal, VP of product management at Informatica, told InfoWorld.

“You can build a wide range of custom AI agents-from employee onboarding assistants and supply chain resilience agents that reroute orders, to marketing personalization agents that coordinate campaigns across CRM and ad platforms,” Pathak said, adding that pre-built templates for common scenarios like chatbots, Q&A bots, data agents, and supervisor/worker patterns among others were available.

The Futurum Group’s lead of data and analytics practice, Bradley Shimmin, said that the agent building capability will be welcomed by enterprises already subscribed to IDMC as it will speed time to market and provide operational support relating to governance and security.

IDC research vice president Stewart Bond said AI Agent Engineering’s agent orchestration is likely to be of more value to enterprises than its agent building capabilities.

“Given that many agents are coming from many vendors across multiple functions, orchestration becomes key. Application and data integration platforms are a natural ‘middleware’ solution for this orchestration,” Bond explained.

AI Agent Engineering is expected to be available globally in fall of 2025.

The adoption challenge

Informatica is far from the only vendor rolling out agentic capabilities for data management.

IBM and SAP have already made some moves towards cleaning up data with agents, said Park.

“Amalgam Insights expects that every data integration company will announce agents this year that assist with data quality, deduplication, lineage, and cleansing — but the devil will be in the details: What percentage of data can the agent support? Can semi-structured and unstructured data be cleansed? Will the agent work with all major enterprise data sources and destinations?” he said.

IDC’s Bond said that getting agents into the market is a start, but helping customers with adoption will be an altogether different challenge.

“These agents will significantly change how people work, and software vendors will also need to be change agents in their customer environments. Training, coaching, and support will be critical to success for both the customer and the vendor,” he said, adding that Informatica will need to invest in new skills in customer success teams and in enabling services partners.

Bond pointed to Informatica’s own Claire Copilot as an example of how long the timeline can be for enterprise adoption.

“We saw Claire Copilot last year at this time, and it is just being made generally available. This demonstrates the reality of how complex these technologies are to work with and package in a way that they are adaptable and scalable in the enterprise,” he said.