Dataiku launches cross-platform AI agent management tool

Dataiku has launched Agent Management, a standalone product for tracking and managing AI agents across enterprise systems. It is aimed at companies running agents on multiple platforms.

The software is designed to give organisations a single inventory of AI agents, including those built outside Dataiku's platform. It also measures business and technical performance and identifies higher-risk deployments.

The launch comes as companies move from testing individual AI agents to operating larger estates of tools across cloud, data and application providers. That shift has created a governance gap: many businesses can closely account for conventional software assets, but not for the AI agents used in their operations.

IBM research cited by Dataiku found that fewer than one in five organisations keep a complete and current inventory of their AI systems. Dataiku argues that most existing agent platforms only provide visibility into agents built within their own environments, leaving companies without a full view of ownership, purpose, cost or risk.

Cross-platform oversight

Agent Management connects to platforms used by enterprise teams, including AWS Bedrock, Databricks Agents, Google Vertex, Microsoft Copilot Studio and Azure Foundry, Salesforce Agentforce, Snowflake Cortex and Dataiku's own software. It also supports OpenTelemetry for custom environments.

The product scans these systems into a single inventory and identifies the structure of each agent, including the models and tools it relies on. The aim is to give managers and governance teams visibility into how an agent works, rather than simply confirming that it exists.

For higher-risk agents, such as those handling customers, sensitive information or live transactions, the system records certification status, named risks and scheduled tests. This creates an evidence trail for managers, auditors and regulators.

The product is positioned above individual vendor stacks rather than within a single platform, allowing companies to compare coverage and risk across a broader set of agents. Users can query the portfolio in plain language to identify unmonitored agents, concentrations of risk and whether agents are justifying their cost.

"Ask a bank how many servers it runs, and you get an answer to the decimal. Ask how many AI agents it's running, and you get a shrug or a guess," said Florian Douetteau, Co-founder and Chief Executive Officer at Dataiku.

"Nobody set out to build it this way. Teams built agents faster than anyone could count them. Agent Management tells you what's actually out there, and what it's actually worth," Douetteau said.

Growing scrutiny

The launch reflects a broader shift in enterprise AI from model experimentation to operational control. As businesses embed AI agents into customer service, internal workflows and transaction-heavy processes, they are facing the same questions long applied to other corporate systems: who owns them, what they cost, how they are tested and what risks they create.

The issue is becoming more pressing as companies adopt services from multiple providers. In many organisations, different teams build or deploy agents through cloud platforms, software vendors or internal systems, making it difficult to maintain a central record.

Dataiku's approach suggests a growing market for tools that sit across competing AI platforms rather than within a single ecosystem. That could appeal to large companies trying to impose consistent oversight over fragmented deployments, particularly where governance requirements span technical, financial and compliance teams.

Commercial model

Agent Management will be generally available in October and priced annually per instance, with monitoring charged per agent. Dataiku did not disclose pricing levels.

Dataiku sells software for building, deploying and governing analytics, machine learning and AI systems across enterprise environments. Its platform is designed to sit above data platforms, cloud infrastructure, applications and AI services, with a focus on centralised oversight in multi-vendor settings.

The release of Agent Management adds a dedicated layer for organisations seeking to catalogue and supervise a growing estate of AI agents spread across several technology stacks.