Databricks Inc. has unveiled significant updates to its Agent Bricks framework, enhancing governance, accuracy, and model flexibility for enterprise AI agents. Announced during the “Week of AI Agents,” new features include expanded capabilities in the MLflow platform for machine learning lifecycle management, a Model Context Protocol (MCP) marketplace, and tools for extracting structured data from documents. These updates facilitate the transition of AI agents from pilot stages to production while ensuring rigorous data access control and decision accuracy.
The MLflow platform will now assess AI agents’ performance, fostering trust and reliability. The introduction of AI Gateway standardizes access to various models, including those from OpenAI and Google. Databricks also launched ai_parse_document, a powerful SQL function for transforming unstructured document data into structured, searchable formats. With these advancements, Databricks aims to support organizations in deploying high-stakes AI agents effectively, pivotal for the future of enterprise AI.
Source link
