BambooAI

BambooAI is a self-hosted AI data analyst for people working with tabular datasets who want to inspect the calculations behind an answer. It turns plain-language questions into Python analysis and charts, then reruns the calculations cited in its report against the original data to check that the numbers reproduce.

The web app runs on your machine, and Docker isolates the code the model writes. It's open source under the MIT license. Ollama and vLLM support local inference on your infrastructure; external providers receive the prompts and analysis material needed for a task. Supported providers include OpenRouter, OpenAI, Anthropic, Google, Groq, Mistral, xAI and DeepSeek. A hosted version runs the analyst and code execution on the service rather than your hardware.

You can inspect each cell's code and output as the analysis proceeds. Follow-up questions retain the previous analysis state, and saved runs preserve notes about the dataset for later work. Quick mode handles short questions, Deep handles fuller analysis, and Adaptive adds a separate model that reviews the work and can redirect it. Reports include figures, uncertainty and a plain-language version, with PDF export available.

An analysis can combine several datasets with supporting papers, notes or transcripts in PDF, Word, Markdown or text format. Sports scientists can load training data directly from Intervals.icu, Endura and SweatStack. With a Gemini key, the analyst can search for published comparison figures and cite those sources in its report.