Dataiku launches tool to inventory and assess enterprise AI agents
Dataiku introduced Agent Management, a standalone product that discovers AI agents across an enterprise regardless of the platform used to build them. It evaluates technical and business performance and highlights agents that carry the most risk. The tool connects to services such as AWS Bedrock, Databricks, Google Vertex, Microsoft Copilot Studio, Salesforce Agentforce, Snowflake Cortex, and Dataiku, with OpenTelemetry support for custom setups.
Dataiku’s new Agent Management is a standalone offering that locates AI agents throughout a company, even those created on different systems. It connects with AWS Bedrock, Databricks Agents, Google Vertex, Microsoft Copilot Studio and Azure Foundry, Salesforce Agentforce, Snowflake Cortex, Dataiku, and OpenTelemetry-based custom environments, then consolidates them into one inventory. It also maps each agent’s tools and models.
For agents tied to customers, sensitive information, or live transactions, the product maintains certification status, identified risks, and scheduled retests, creating an audit-ready record. IBM research cited in the announcement found fewer than 20% of organizations possess a full and current count of their AI systems. Most agent platforms only surface agents built on themselves.
Tools such as Agent Management could affect IT, security, compliance, and audit teams by making enterprise AI agents more visible and comparable. Customers and workers may benefit if risky or poorly performing agents are identified sooner, though broader monitoring may raise privacy and oversight questions. The actual societal effect may depend on adoption, governance, and whether organizations use the findings to improve accountability rather than simply expand surveillance.