Dataiku Ships Standalone Agent Governance With a Bet That Cross-Platform Beats Native

Dataiku announced Dataiku Agent Management on March 9, 2026, a standalone cross-platform agent governance product that connects to agents built on Microsoft Copilot Studio, Azure Foundry, Salesforce Agentforce, AWS Bedrock, Google Vertex, Databricks, Snowflake Cortex, n8n, and Dataiku itself without requiring the Dataiku platform. The product discovers agents into a single inventory, measures them against business KPIs for usage, cost, quality, and value, and certifies them for risk with scheduled tests and audit-ready records. Dataiku CEO Florian Douetteau told CIO Dive the company is working with customers on a central access point for agents, citing Gartner's projection that the average Fortune 500 enterprise will manage more than 150,000 AI agents by 2028, up from fewer than 15 in 2025.

Most enterprise agent governance ships inside the platform that built the agent. Salesforce has its Einstein Trust Layer. Microsoft has Agent 365 paired with Purview. Each one is strong inside its own walls and silent outside them. Dataiku is betting there is room for something that sits above all of those walls at once. Dataiku Agent Management https://www.dataiku.com/product/agent-management , announced March 9, 2026, as part of the company’s Platform for AI Success launch, is positioned as a standalone cross-platform control tower. It connects to agents built on Microsoft Copilot Studio, Azure Foundry, Salesforce Agentforce, AWS Bedrock, Google Vertex, Databricks, Snowflake Cortex, n8n, and Dataiku itself. It does not require the Dataiku platform — a meaningful departure for a company whose core business is an integrated data science environment. The product does three things. First, it discovers agents across platforms into a single inventory, capturing who owns each agent, what it is supposed to do, and what data it can access. Second, it measures agents against business key performance indicators — usage, cost, quality, and value — rather than just uptime. Third, it certifies agents for risk, runs tests on a schedule, and maintains audit-ready records. “Technical monitoring tells you agents are running,” the company notes in its enterprise agent systems documentation https://www.dataiku.com/blog/enterprise-agent-systems . “It does not tell you they are doing their jobs.” The distinction matters because the scale of the agent fleet problem is accelerating. Gartner projects the average Fortune 500 enterprise will manage more than 150,000 AI agents by 2028 — up from fewer than 15 in 2025. Yet only 18% of organizations maintain a complete, current agent inventory, and less than half have visibility into what agents exist across their environments, according to IBM Institute for Business Value https://www.ibm.com/thought-leadership/institute-business-value/en-us/report/2026-cxo research. An OutSystems https://www.outsystems.com/1/state-ai-development/ 2026 survey of 1,900 IT leaders found that 96% of enterprises run AI agents but only about 12% have a centralized way to manage them. McKinsey, citing a SailPoint survey, found that 80% of enterprises report risky behavior from live agents. The infrastructure is lagging the deployment. Dataiku’s framing https://www.dataiku.com/blog/enterprise-building-agents-account is direct: most enterprises can build agents but cannot account for them. The company’s pitch is that platform-native governance — Salesforce’s Trust Layer for Agentforce agents, Microsoft’s Agent 365 and Purview for Copilot agents — creates a patchwork of siloed visibility. Each tool covers its own territory. None of them see the full fleet. CEO Florian Douetteau told CIO Dive that the company is working with customers on a central access point for agents, identifying a growing appetite for lifecycle management across fragmented deployments. The timing connects to a broader pattern. Four vendors shipped standalone agent governance tools /the-agent-governance-stack-is-forming-four-products-two-weeks-one-pattern/ in a 14-day window between August 24 and early September — Okta https://www.okta.com/newsroom/press-releases/okta-brings-first-class-identity-to-ai-agents-with-agent-sso/ with Agent SSO, IBM https://www.ibm.com/new/announcements/new-in-ibm-watsonx-orchestrate-cross-platform-agent-discovery-custom-evaluation-and-agentops-agent-goes-ga with watsonx Orchestrate AgentOps Agent, Broadcom with AgentMinder, and Dataiku. Each approached from a different infrastructure layer: identity, orchestration, security, and observability. Dataiku’s angle is business-value measurement — not just whether the agent is secure or running, but whether it is delivering results that justify its existence. The competition is not just other point tools. It is the native governance that every major cloud and SaaS vendor already bundles. Salesforce’s Einstein Trust Layer provides PII masking, data redaction, toxicity filtering, and retention controls scoped to Salesforce’s environment. Microsoft’s Agent 365 manages agent identity and lifecycle at the tenant level, while Purview adds data loss prevention and sensitivity labels across Microsoft’s stack. Both are deep but proprietary. Dataiku is wagering that enough enterprises have agents spread across enough platforms that the cross-platform view has more value than the deep-but-narrow native one. There is one wrinkle worth noting. Dataiku’s initial press release pointed to a September 2026 general availability date. The current product page lists October 2026. Dataiku’s enterprise agent systems blog https://www.dataiku.com/blog/enterprise-agent-systems from May 2026 references a September launch with early access available now. The product page FAQ says October. For a product whose pitch is reliability and auditability, the moving date is a minor signal, but it is worth flagging. Enterprise buyers evaluating governance tools should verify the GA timeline directly with Dataiku before mapping it into their own deployment plans. The real question is not whether Dataiku’s product works — it is whether enterprises are ready to add another vendor to manage the vendors they already have. Agent governance is a genuine need. But governance tool sprawl — managing the tools that manage the agents — is its own emerging problem. Dataiku is selling a single pane of glass over a heterogeneous fleet. Whether that pane is clearer than the collection of native dashboards it replaces depends on how many platforms an enterprise actually runs, and whether the overhead of a cross-platform layer is worth the unified view.