Ryoma: Building Open Source AI Agents for Enterprise Data Analysis
summary
Ryoma is an open source AI agent framework that translates natural language into SQL, runs safety checks and prompt validation before calling models, and connects to Snowflake, Postgres and Spark using LangChain-based SQL tools and a workflow-as-graph agent runtime. It pre-indexes metadata into a retrieval layer with PGVector (or Milvus/others), generates multiple SQL candidates for voting, supports terminal, web (Python Reflex) and notebook UIs, reports adoption by JPMorgan PayPal and TikTok yet remains early with ~300 stars.
AI-written summary · about
chapters (8)
- 0:00Ryoma and open source AI agents
- 3:13High-level architecture
- 5:04Core framework and workflows
- 9:28Tooling and SQL safety
- 13:47React system and data catalog
- 18:39Current infrastructure and frontend
- 22:17Future roadmap
- 25:32Live demo: terminal
related talks
15mGraphRAG-Powered AI Agents: Transforming Corporate Metadata into Intelligent Knowledge
38mScaling Specialist Knowledge with AI
45mForget SQL! Create your own AI Data Analyst
01h 01mBuilding the Intelligent Enterprise: Ensuring Success with Data and GenAI
30mBoosting Sales with Intelligent BI Agents
40mRolling out Enterprise AI: Tools, Insights, & Team Empowerment
19mDatabricks
24mBuilding Your Intelligent Data Agent