Dataiku debuts cross-platform Agent Management, expands Cobuild building agent
SiliconANGLE Duncan Riley ● Covered by 2 sources
Dataiku launched a tool to track every AI agent a company runs, even from other vendors. It also widened Cobuild, so people can build and check agents in plain English.
Based on reporting by SiliconANGLE, Duncan Riley — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Dataiku is trying to solve a mess that most enterprises still haven’t cleaned up: they can count servers and software licenses, but not their AI agents. On Tuesday, the company launched Agent Management, a standalone product that pulls those agents into one inventory even when they live on other vendors’ platforms. It also expanded Cobuild, its natural-language builder for AI projects.
Agent Management connects to Salesforce Agentforce plus agent services from Amazon Web Services, Microsoft and Google. It also covers agents built on Databricks and Snowflake, Dataiku’s own agents, and custom environments that report through OpenTelemetry. Once those systems are connected, the product maps what each agent depends on — tools, models and other pieces — so teams can see how the thing actually works instead of guessing.
That matters because Dataiku says most companies are flying blind. The company cited IBM research showing fewer than one in five organizations keep a complete and current inventory of their AI systems. Florian Douetteau, Dataiku’s co-founder and chief executive, put the contrast bluntly: a bank can usually say how many servers it has “to the decimal,” but ask about AI agents and the answer is “a shrug or a guess.”
The product is also meant to handle the unpleasant parts. Riskier agents get a standing record of certification status and named risks, and tests rerun on a schedule so an evidence trail exists before an auditor or regulator shows up. Dataiku is pitching that against monitoring tools embedded in a single vendor’s stack, which it says naturally favor that vendor’s own agents. Sitting above those stacks lets it answer broader questions, like where risk is concentrated or which agents cost more than they return.
Cobuild is moving in the same direction. Dataiku made it generally available in June, and now it reaches farther across the platform. Cobuild Insights lets employees ask questions of governed company data in plain language, while Dataiku Headless brings Cobuild into developer tools such as Claude Code, Codex and Cursor. The company also says old Excel workbooks or Alteryx workflows can be turned into governed projects, or pointed at a live environment to investigate problems. Agent Management will be generally available in October, with annual per-instance pricing and monitoring charged by the agent.
My take — AI-written commentary, not fact-checked reporting
This is the unglamorous part of AI that actually matters: inventory, controls and blame assignment. Everyone loves talking about agents until someone asks who owns them, what they touch, and why they’re still running after the demo. The real product here isn’t magic AI; it’s corporate memory, which turns out to be rarer than the agents themselves.
Read more about this at: SiliconANGLE
