CData and Komprise launch AI agent governance tools amid security fears

CData and Komprise have each rolled out tools designed to govern how AI agents interact with enterprise data, entering a space that has become increasingly urgent as deployments scale and security concerns grow. Both launches target the Model Context Protocol, an open-source framework that lets agents query external data sources, but which has also raised questions about unintended access.

CData’s Connect AI Gateway, now in early access, extends the company’s existing AI platform to let IT teams set policies on model selection, tool usage, and data access. The product is positioned as a way to help prevent agents from operating outside their intended parameters—an issue that has drawn attention in recent discussions about enterprise AI risks. Komprise, meanwhile, is addressing what it calls “MCP bloat” with its Universal File MCP tool, which consolidates disparate data sources behind a single interface to simplify queries and reduce potential misconfigurations.

The timing of these launches aligns with broader industry trends. Just days ago, ServiceNow made its own push to position itself as the default “control tower” for AI agents, framing governance as a balance between risk and business value. That conversation has intensified as enterprises grapple with how to manage AI agents at scale. While neither CData nor Komprise referenced specific incidents directly, both products reflect a growing focus on the challenges of securing agent-data interactions.

What sets these offerings apart is their focus on the MCP layer, rather than the broader agent lifecycle. Collibra, which we covered last week, rolled out runtime governance tools that monitor agents in real time, while ServiceNow is pitching a more holistic framework for incident response. CData and Komprise, by contrast, are zeroing in on the specific problem of data access—a critical concern for most enterprise deployments. That suggests a market bifurcating: some vendors are building guardrails for the entire agent workflow, while others are specializing in the most immediate pain points.

The question now is whether enterprises will adopt these tools proactively or react to emerging challenges. The latter seems more likely, given the pace at which AI agents are being deployed relative to governance infrastructure. The demand for compute and tooling, as seen in recent coverage of providers like CoreWeave, shows momentum behind AI adoption, but security and compliance remain secondary considerations for many teams. That dynamic may not hold if enterprises face more visible consequences.

For now, CData and Komprise are betting that their technical solutions will address practical needs ahead of policy shifts. Whether they succeed may depend less on the quality of their products and more on how quickly enterprises recognize the risks of waiting.

Sources: siliconangle.com · siliconangle.com