For two years the AI conversation in the enterprise was about answers. Ask a model a question, get a paragraph back. Useful, but it does not touch the systems where work actually happens. The conversation has now moved to agents, software that does the work, not just describes it. And an agent is only as good as the tools it can reach.
So we made Nexadata one of those tools. Our latest release ships a Model Context Protocol server.
What that actually means
Model Context Protocol is becoming the common language between AI agents and the systems they act on. By speaking it, Nexadata stops being something a person clicks through and becomes something an agent can call directly. An agent can discover the workflows and pipelines you have built, run them, pass in values at runtime, and read back execution telemetry to see what happened.
The release has three parts that make that real.
- The MCP Server exposes your workflows, pipelines, and execution telemetry to agentic AI clients.
- Runtime variables let an external agent inject values when it triggers a workflow, so one pipeline serves many requests.
- Execution statistics give per-run metrics, in the UI and through the server, so an agent (and a person) can see exactly what a run did.
Why this is the important one
Here is the part I care about. When an agent calls Nexadata, the same boundary still holds. The agent describes intent. Our engine executes deterministically, with lineage and a human review step where it matters, and raw data stays inside the platform. So you get the reach of agentic AI without giving up governance or handing your data to a model.
That is the whole bet. The world is going to fill up with agents that want to do real work on real data. The ones that matter will be the ones you can trust to touch production systems. We built the layer that lets them do that safely.
More on the architecture on the platform page, and on the boundary on the security page.