Report Abuse

Basic Information

BambooAI is an open-source Python library and multi-agent system designed to let users interact with tabular and external data using natural language. It provides conversational data analysis by classifying user questions, routing them to specialized agents, generating Python analysis and visualization code, executing that code, and returning formatted results. The project supports both interactive use in Jupyter/CLI and a web UI, and can be run locally, with Docker, or as a pip package. It is intended for analysts and developers who want to augment or automate data exploration, cross-reference auxiliary datasets, integrate web search and APIs, and persist successful solutions in a vector store. Configuration is driven by an LLM_CONFIG.json and environment variables for model and service credentials.

Links

Categorization

App Details

Features
Natural language interface for data analysis with conversational and single-query modes. Multi-agent architecture with configurable agent roles such as Expert Selector, Planner, Code Generator, Error Corrector, Reviewer and Solution Summarizer. Automatic dynamic prompt building, code generation, execution and LLM-based debugging and retry loops. Support for local and remote models, multiple providers, Docker deployment, pip package and full repository deployments. Optional internet search tooling, auxiliary dataset handling, dataframe ontology support for semantic grounding, and vector database integration for episodic memory and solution reuse. Web UI and Jupyter notebook support, logging of runs and token/cost metrics, custom prompt templates and extensive model configuration via LLM_CONFIG.json.
Use Cases
BambooAI lowers the barrier to data analysis by converting plain language questions into executable Python code and visualizations, enabling users without deep coding expertise to obtain insights. The system guides complex tasks through planning agents, requests clarifying feedback when prompts are ambiguous, and self-corrects generated code using dedicated error-correction agents. It can combine primary data with auxiliary datasets, use ontologies to improve semantic understanding, and store high-quality solutions in a vector DB for future retrieval, improving productivity over time. Deployable as a web app or in notebooks, it helps teams prototype analyses, reproduce workflows, and evaluate model performance across providers while providing logs and configuration for auditability.

Please fill the required fields*