Databricks Launches New Tools for Scalable and Governed AI Agents

Databricks introduced new tools on Tuesday to help enterprises scale AI agents from pilot projects to full production while ensuring governance, monitoring, and integration. These tools, which include Mosaic AI Gateway, AI/BI Genie Conversational API, Agent Evaluation Review App, and Batch AI, are currently available in public preview.

The new tools address challenges enterprises face in deploying AI agents for high-value use cases. According to Databricks, 85% of global enterprises use generative AI, but models often fail to deliver accurate and well-governed outputs due to a lack of enterprise data awareness.

“Many enterprises still struggle to deploy AI agents in high-value use cases due to concerns around accuracy, governance, and security. For these organisations, it’s confidence, not just technology, that presents the biggest hurdle to extracting the full data intelligence benefits of generative AI,” Craig Wiley, senior director of product for AI/ML at Databricks, said. “The new tools address these challenges head-on, enabling businesses to move beyond pilots and into full-scale production with AI agents they can trust.”

The Mosaic AI Gateway provides centralised governance by integrating and managing open-source and commercial AI models within a single platform. It supports custom large language model (LLM) providers and ensures unified governance, monitoring, and integration across models. 

Moreover, the AI/BI Genie Conversational API allows developers to embed AI-powered chatbots into custom applications and productivity tools such as Microsoft Teams, SharePoint, and Slack. The API retains context across conversations, allowing for follow-up queries without loss of continuity.

The Agent Evaluation Review App streamlines human-in-the-loop workflows, allowing domain experts to provide structured feedback, send traces for labelling, and customise evaluation criteria. This eliminates the need for spreadsheets or custom-built applications. 

The provision-less batch inference tool allows enterprises to run batch inference using a single SQL query, integrating with Mosaic AI without requiring infrastructure provisioning. Earlier this year, Databricks closed $15.3 billion in financing, valuing the company at $62 billion. JPMorgan Chase, Barclays, Citi, Goldman Sachs, and Morgan Stanley led the financing, which included $10 billion in Series J equity funding and $5.25 billion in debt financing.