Dataiku Agent Management: The Cross-Platform Governance Layer for Enterprise AI Agent Sprawl

Dataiku CEO Florian Douetteau described the problem in one line: ask a bank how many servers it runs and you get an exact number. Ask how many AI agents it runs and you get a shrug or a guess.

On September 24, 2026, Dataiku announced Agent Management, a standalone product designed to solve that problem. It is one of four standalone agent governance tools shipped in a 14-day window between late August and early September 2026, alongside Okta Agent SSO, IBM watsonx Orchestrate AgentOps, and Broadcom AgentMinder. Each approached agent governance from a different infrastructure layer: identity, orchestration, security, and observability. Dataiku's angle is business-value measurement.

The product reaches general availability in October 2026, priced per instance per year with metered monitoring per agent. It does not require the Dataiku platform, a meaningful departure for a company whose core business is an integrated data science environment.

What Dataiku Agent Management Actually Does

The product does three things:

1. Discovery and Inventory

Agent Management 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. It connects to nine platforms: Microsoft Copilot Studio, Azure Foundry, Salesforce Agentforce, AWS Bedrock, Google Vertex, Databricks, Snowflake Cortex, n8n, and Dataiku itself. OpenTelemetry support covers custom environments.

2. Business KPI Measurement

The product measures agents against business key performance indicators: usage, cost, quality, and value, rather than just uptime. Dataiku's own documentation makes the distinction: "Technical monitoring tells you agents are running. It does not tell you they are doing their jobs."

3. Risk Certification and Audit

Agent Management certifies agents for risk, runs tests on a schedule, and maintains audit-ready records. Each inventory entry can include the tools and models an agent uses, its owner and purpose, usage and cost, plus risk assessments and certification status for higher-risk agents.

The Scale of the Agent Sprawl Problem

The numbers behind this product explain why it exists:

The infrastructure is lagging the deployment. Teams build agents inside whichever platform is closest: a Copilot Studio agent in one department, a Bedrock agent in another, a Salesforce agent owned by sales operations. Each platform sees its own agents. Nobody sees all of them.

What Agent Management Does Not Do

The limits matter as much as the capabilities. According to eWeek, Dataiku's tool is not an AI gateway sitting between agents and their tools. It does not see real-time actions and cannot stop an agent mid-task. Dataiku's own documentation states that integration depth depends on what each connected platform exposes.

This is a real distinction. An inventory tells you which agents exist, who owns them, and how risky they look. Runtime control, meaning permissions, rate limits, and kill switches, sits elsewhere. Buying one and assuming you have the other is an easy mistake.

Cross-Platform Governance vs. Native Dashboards

The core question Dataiku raises: is a cross-platform view more valuable than the deep-but-narrow governance that each platform already provides?

The Case for Native Governance

Major cloud and SaaS vendors already bundle agent governance:

• Salesforce Einstein Trust Layer: PII masking, data redaction, toxicity filtering, retention controls scoped to Salesforce's environment.
• Microsoft Agent 365 + Purview: Agent identity and lifecycle management at the tenant level, with data loss prevention and sensitivity labels across Microsoft's stack.

Both are deep but proprietary. They work well inside their own walls and are silent outside them.

The Case for Cross-Platform

Dataiku's argument is that enough enterprises have agents spread across enough platforms that the cross-platform view has more value than the deep-but-narrow native one. With two platforms, native tools may be enough. With five, a spreadsheet stops working.

The counterpoint, raised by Forkast, is fair: adding a vendor to watch your other vendors is its own overhead. Whether that overhead is worth the unified view depends on how many platforms an enterprise actually runs.

The Governance Stack Is Forming

Dataiku Agent Management is part of a broader pattern. Four vendors shipped standalone agent governance tools in two weeks:

This suggests the agent governance stack is forming as a distinct product category. Each vendor approaches from a different infrastructure layer, but the problem is the same: enterprises cannot manage what they cannot see.

Gartner's six-step framework for agent governance places a centralized agent inventory as step two, right after governance policy. The implication is clear: you cannot enforce much until you know what exists.

Practical Takeaways for Enterprise Leaders

Start With a Minimal Record

You can begin before any purchase. For every agent, capture:

1. A named human owner and the business purpose
2. The platform it runs on and the models and tools it can call
3. The data it can read and the actions it can take
4. Risk tier, last review date, and monthly cost

If an agent has no owner, that is your first finding.

Decide Who Owns the List

Servers got asset management and cloud accounts got cost tooling once counts got out of hand. Agents are following the same path, only faster. Whether you choose Dataiku, a native dashboard, or a well-kept register, decide who owns the list before the count reaches four digits.

Know the Limits

An inventory is not runtime control. Knowing which agents exist and how risky they look is different from being able to stop one mid-action. If you need kill switches, rate limits, and permission enforcement, you need a gateway, not just an inventory.

Read Next

• ZoomInfo Agent Teams: The First GTM Platform to Bundle Multi-Agent Orchestration
• The $74.3B AI MarTech Market: What Tripled Marketing Tech Spending Means
• IBM Bob Goes Self-Hosted: Why AI Sovereignty Just Became an Enterprise Requirement
• WSO2 Agent Manager: Open-Source Governance for AI Agent Sprawl

Need help building your AI agent governance framework? Contact Vivid to design your agent inventory and controls.