Surus Agentic Postgres Companion (github.com)

The announcement of Surus, a Postgres companion client featuring a built-in agent, represents a significant advancement in data querying for the AI/ML community. Surus is designed with three core principles: a tight text-to-SQL loop for optimized queries, a read-only architecture for safety, and a default focus on data privacy. This ensures that while leveraging large language models (LLMs) for generating precise SQL queries, users can trust that their production databases remain secure and unaffected by unintended modifications. Surus offers a robust environment for interacting with databases, integrating modern SQL editing capabilities with an intelligent agent that transforms plain text requests into optimized queries. This includes a unique three-mode chat interface (SQL, Question, Teach) and an "Explain with AI" feature to demystify SQL execution plans. Additionally, it provides valuable functionalities such as auto-generated Entity-Relationship Diagrams (ERDs), detailed query logging, and a user-friendly interface that can operate both as a local web app and a native MacOS app. These developments hold promise for streamlining database interactions, improving query efficiency, and enhancing user comprehension of complex database structures, making it a notable tool for data professionals and AI enthusiasts alike.