Databricks Launches New Solutions for Better Implementation and Governance of AI Agents
Databricks expands the Data Intelligence Platform with new features to develop and manage AI agents more efficiently. The improvements include easier integration with applications, centralized governance, and enhanced human control over AI models.
Integration and Management
One of the new capabilities is the integration of AI/BI Genie via API with the Mosaic AI Agent Framework. This allows companies to obtain data insights through chatbots on platforms such as Slack, Teams, and SharePoint. This helps organizations to connect data and solve domain-specific challenges.
read also
Databricks introduces new tools for AI agents and governance
Additionally, the Mosaic AI Gateway now also supports self-developed LLM models. This allows companies to manage both commercial and open source AI models in one place. They can also make use of Databricks features such as AI Playground and inference tables.
To further improve the quality of AI models, the Agent Evaluation Review application has been adapted. This makes it easier for employees from different departments to provide feedback on AI models.
Batch Inference
Another new feature is provisionless batch inference. This allows large amounts of data to be processed via a single SQL query, without the need to manually set up infrastructure. This reduces the operational burden and allows teams to focus entirely on scaling their AI projects.
According to Databricks, it is becoming increasingly important not only to apply AI but also to scale it effectively and in a structured manner. The new features help organizations to deploy AI more broadly while ensuring governance and accuracy. The new capabilities are available as a public preview.