Dataiku’s platform now allows companies to create and manage AI agents at scale. The solution combines analytics, predictive models, and agents within a single enterprise platform. This gives companies more control over their AI applications and allows them to integrate them with existing data systems.
According to Dataiku, many companies that implement AI agents face problems. Agents operate without proper IT control, vary in quality and relevance, and are spread across different teams without management. Dataiku aims to solve this by enriching the platform with agents as fully-fledged business systems – based on reliable data, embedded in operational workflows, and managed with the same precision as other business-critical resources.
Dataiku now supports visual agents (no-code for non-technical business users) and code agents (full-code for developers). Both are available on the same platform with essential governance features.
Agentic ecosystems bring security and management risks, especially as models evolve rapidly and costs remain difficult to monitor. Dataiku therefore offers new possibilities for IT departments to orchestrate agents at scale. For example, there is the Dataiku LLM Mesh for managing model access between all providers. In addition, the platform features Dataiku Safe Guard, which allows flexible security measures to be defined and applied. Finally, Agent Connect is a tool that provides access to all agents within the organization from a single interface.
Monitoring for continuous optimization
Dataiku also provides a continuous optimization approach with features such as Trace Explorer, which provides complete transparency in decision-making, Quality Guard, which continuously evaluates agent performance, and Cost Guard, which provides real-time cost tracking and budget enforcement.
With an agnostic approach, Dataiku supports all major cloud environments, model vendors, and data platforms, including Snowflake, Databricks, Microsoft, AWS, and Google. This enables customers to integrate agents into existing data flows, MLOps workflows, and model governance processes, eliminating duplication and achieving higher returns.
The new agentic AI functionality is available immediately.