CData Software Inc., the maker of a platform and drivers that connect data from multiple sources to applications, today announced a version of its product designed for artificial intelligence assistants, workflow builders and agent frameworks.

The company said ConnectAI provides real-time, governed access to more than 300 enterprise data sources using the Model Context Protocol, a standard framework that lets AI systems securely connect to and interact with external data sources and applications.

“It’s a single platform that allows you to establish connections to and manage the enterprise sources that you have across your business, whether [large language model], agentic platform or workflow platform,” said Will Davis, CData’s chief marketing officer. “The focus is taking this connectivity that we’ve built and evolving it to meet the needs of where organizations are going with AI.”

CData, which has raised over $500 million in venture capital and private equity funding, provides real-time access to more than 300 applications, databases and web application programming interfaces with a self-service approach that resolves format variances and allows applications to read and write to multiple data sources simultaneously.

The company built on its existing technology to create ConnectAI, said Chief Product Officer Manish Patel. “Enterprises need to have a centralized, governed way to manage connectivity across agents; they’re not just going to use open source or a variety of connectors in the background,” he said.

He likened MCP to “USB-C for connectivity for AI,” allowing platforms like Anthropic pbc’s Claude, the open-source LangChain LLM framework and OpenAI LLC’s ChatGPT to access enterprise systems without custom integrations.

CData’s connectors standardize communication using SQL-92, regardless of whether the underlying data source supports SQL. “You can speak one language to any one of those connectors, and we basically convert that SQL dialect into what the [application programming interface] understands,” Patel said.

CData made performance optimization a priority when developing ConnectAI, Patel said. “If you just give an LLM unfettered access to an API, it’s going to pull a huge data set just to do an order-by,” he said. “We understand how much of a query you can push down to the server side and how much we have to do client side, so we are minimizing the token consumption as much as we can.” The product was tested with more than 1,600 users.

The company plans to expand Connect AI’s capabilities early next year by offering customization capabilities that enable customers to limit agent access to specific tools. Connect AI is available immediately.

Image: SiliconANGLE/DALL·E