Dataiku Inc., the creator of a data science platform for artificial intelligence developers, is beefing up the agentic capabilities of its Universal AI Platform with the launch of a new Agent Hub.

It’s part of an ongoing effort to help companies maintain better control over AI agents, which operate autonomously with limited human oversight. In the rush to embrace AI agent automation, many companies have deployed agents with clumsy architectures that aren’t always reliable and operate outside the control of information technology teams, introducing a huge amount of risk.

Dataiku’s Universal AI Platform allows companies to deploy AI agents on a much safer footing, within a centralized portal that ensures they’re grounded in trusted data, so they can be embedded in operational workflows without problems and governed by the same rules as traditional enterprise systems.

The company launched its agentic development tools earlier this year, and the Agent Hub builds on that, providing a centralized workspace for employees to discover, use and also create “approved” AI agents that remain under the full control of IT teams. It’s a trusted home for every kind of AI agent, including those built by third-parties and companies’ own development teams.

The main advantage of the Agent Hub is that it eliminates the problems caused by having numerous ungoverned agents getting access to sensitive data and playing around with critical systems without any oversight. Instead, IT teams must approve each new agent individually before employees can access them. They can also determine which underlying large language models power each agent, as well as the data they can access. By deploying AI agents in Dataiku’s Agent Hub, businesses also benefit from native integration into data pipelines and machine learning workflows, eliminating much of the grunt work that’s involved in their creation.

Dataiku said the Agent Hub also comes with more agentic development tools, including a new “simple builder” that anyone can use to create AI agents for common workflows, based on a library of more than 20 templates. In addition, IT teams can keep track of who is using each AI agent, its performance and other metrics, so they can identify which ones are having the most positive impact on the business.

Co-founder and Chief Executive Florian Douetteau said the launch of the Agent Hub will help put an end to the era of isolated experiments and limited oversight. Instead, companies can now access enterprise-grade AI agents with strict controls and start deriving real value from their automation efforts.

Image: SiliconANGLE/Meta AI