CData Software is expanding the connectivity, context and control capabilities of its CData Connect AI platform to improve data accuracy for AI applications and agents in production.
CData is expanding the capabilities of its CData Connect AI platform, the company’s MCP (Model Context Protocol) server, to improve links to enterprise data for AI applications as they move from development into production.
The platform’s new connectivity, context and control capabilities are designed to address what CData describes as data infrastructure gaps, including problems with data access and accuracy, that hinder AI applications and agents when they reach the production stage.
CData executives cited the company’s own market research that found that 53 percent of organizations rely on custom-built APIs, connectors and data pipelines to support enterprise AI, while 71 percent said AI teams spend more than a quarter of implementation time on data integration chores.
[Related: Meeting The Data Needs Of The AI World: The 2025 CRN Big Data 100]
“With everything that’s going on in AI there’s a huge need around how you safely perform—in a high-performance manner and also high-accuracy manner— [and] connect enterprise data sources to the different AI technologies that organizations are utilizing internally,” said Will Davis, CData chief marketing officer, in an interview with CRN.
“The need to be able to connect enterprise data to those AI systems and do so with the right control, the right context and [in] a highly performing manner, and also make sure that those answers are accurate, is a pretty challenging area for a lot of organizations,” he said.
The new capabilities extend the functionality of the CData Connect AI platform that launched in September. Connect AI, which builds on the company’s flagship CData Connectivity Platform, integrates AI applications, agents and workflows with 350-plus enterprise data sources, such as databases and operational applications, providing AI systems with the governed, real-time business data they need to operate effectively.
While there are many data connectivity and MCP tools on the market, CData executives said the company’s goal is to position CData Connect AI as an independent data infrastructure layer that promises flexibility and data accuracy for AI.
Update Details
On the connectivity side the platform now supports links to more than 350 business systems, up from around 300 when the product initially debuted. Also new is Connect Gateway that extends the reach of Connect AI to data sources behind a firewall, such as SAP applications, SQL server and PostgreSQL databases, and others.
For data context, the platform now offers expanded agent tooling and toolkits. New universal tools, for example, provide a normalized set of operations that work consistently across all 350-plus connected systems, while new source tools expose tightly defined operations specific to each system. And new custom tools provide a way for organizations to define purpose-built operations for specific workflows by executing pre-optimized queries with explicit data access limits.
For control, the platform now includes data governance enhancements including SCIM (System for Cross-domain Identity Management) 2.0 for automated identity lifecycle management, along with Custom OAuth Applications that enable organizations to use first-party credentials to meet internal security and compliance requirements, with every query being authenticated, authorized and auditable.
CData tested Connect AI against other MCP platforms for establishing AI agent connections with data sources, including enterprise and CRM applications, a project management application and a data warehouse, and said Connect AI performed with 98.5 percent accuracy compared to accuracy results of 65 to 75 percent for other MCP systems.
The latter accuracy rate is far too low for an agentic AI system to be effective, said Jerod Johnson, CData director of technology evangelism, in the interview with CRN.
“If you want something to run autonomously, you need extremely high accuracy,” he said. And he noted that accuracy rates drop even lower over multi-step AI workflows as errors compound.
CData works with a number of global and regional system integrators, some of whom have practices focused on assessing the “AI readiness” of clients’ data infrastructure, CMO Davis said. And many assemble pre-architected AI technology stacks for clients that incorporate the CData software.
“This is a top initiative for any organization that’s investing in AI today,” he said.
