Dataiku Launches Cross-Platform AI Agent Management

Here is a number that should give any IT leader or operations executive pause: fewer than one in five organizations maintain a complete, current inventory of their AI systems. That is not a security researcher’s projection. That is from IBM’s “AI in Motion” research, published while most of those same organizations were actively scaling up their AI agent programs.

The gap between how fast companies are deploying AI agents and how well they can account for them is one of the more quietly dangerous problems in enterprise technology right now. Dataiku is trying to close it.

On September 24, 2026, Dataiku launched Agent Management, a standalone product built to find every AI agent running in your enterprise, regardless of which platform built it, and give you a clear picture of what each one is doing, how well it is performing, and how much risk it carries.

What the Product Does

Dataiku Agent Management works across the major enterprise AI platforms. It inventories agents running on AWS, Databricks, Google Cloud, Microsoft Azure, Salesforce, and Snowflake, alongside any agents built directly within Dataiku itself. The product does not require those agents to be rebuilt or migrated. It connects to where they already live.

From there, it does three things:

Inventory. It finds agents you may not even know exist. Large enterprises often have dozens of AI projects running in different business units, built by different teams, on different platforms. Agent Management creates a single, current list.

Performance tracking. The product measures agents against business KPIs, not just technical uptime metrics. Whether an agent is technically running without errors is a different question from whether it is delivering the business result it was deployed to produce. Agent Management tracks both, and the distinction matters.

Risk tiering. Not all agents carry the same risk. An agent summarizing internal meeting notes operates in a different risk category than one processing payment authorizations or making procurement decisions. Agent Management tiers agents by their risk profile so that governance resources can be applied proportionally.

The product will be generally available in October 2026. Pricing is per instance annually, with monitoring metered per agent.

Why This Matters

The problem Dataiku is solving is structural, not technical. Enterprises have not been slow to adopt AI agents. In many cases, they have adopted them faster than their governance and visibility frameworks could keep up. The result is an environment where business units are deploying agents independently, central IT does not have full visibility, and the organization as a whole cannot answer basic questions like: how many agents do we have running, are they working, and what are the consequences if they behave unexpectedly?

That is a risk posture most companies would find unacceptable with any other class of business software. The reason it has been tolerated with AI agents is that until now there was no clean solution that worked across the multi-vendor reality of how enterprises actually deploy AI.

The timing reflects a broader maturation in the market. Gartner is forecasting that 40 percent of enterprise applications will embed task-specific AI agents by the end of 2026, up from less than 5 percent in 2025. At that adoption rate, governance tools stop being nice-to-have and become operational necessities.

What This Means for Business

If your organization is running more than a handful of AI agents, the questions Agent Management answers are ones you should already be asking: Who owns each agent? What is it responsible for? Is it performing? What happens if it fails or behaves outside its intended scope?

The fact that most organizations cannot answer those questions is not a failing of any one team. It reflects how quickly the agent landscape has moved and how fragmented the tooling has been. A product like Agent Management starts to address that by creating a governance layer that works across the platforms businesses are already using.

There are a few practical implications worth considering:

Auditability is becoming a real requirement. Regulatory scrutiny of AI systems is increasing, and the question of whether a company can account for its AI agents is starting to appear in procurement questionnaires, insurance evaluations, and compliance reviews. Having an inventory is not optional in that environment.

Performance data needs to be business-level. Many teams evaluate their agents based on technical metrics: latency, error rates, uptime. Those metrics tell you if the system is running. They do not tell you if the business outcome the agent was deployed to produce is actually happening. Closing that gap is where ROI from AI agents is either realized or lost.

Cross-platform reality requires cross-platform tools. Enterprises rarely run their entire AI stack on a single vendor’s infrastructure. Governance tools that only work within one platform’s ecosystem will always leave blind spots. This is a structural problem, and Agent Management’s approach of connecting to agents wherever they run is the right architectural response.

For organizations scaling their AI agent deployments, the ability to see everything in one place, assess performance against real business outcomes, and manage risk proportionally is not a feature request. It is a precondition for responsible deployment at scale.

---

Enterprise DNA builds and governs AI agent systems for businesses across operations, finance, customer service, and more. If you are navigating the challenge of deploying AI agents at scale, speak with our team about what a properly governed deployment looks like for your business.