I'm Vahid, and excited to share DataBot, a self-hosted data assistant that lets you query, visualize, analyze, and forecast data in plain language.

Backstory: Right after ChatGPT launched, I built a prototype chatbot for Metabase (a BI tool) to answer user questions with natural language. That experiment grew into an independent app — now a fully-fledged AI-powered data assistant.

What's different:

- Self-hosted: Full isolation and control over your data. No SaaS vendor, no third-party access.

- Feature-rich: Includes forecasting, NoSQL (MongoDB), JSON column support, Metabase integration, and data obfuscation — things rarely found in other tools.

- Affordable: Free for personal use. Business plans start at $12/user/month.

Technical details:

- Framework: FastAPI

- Data explorations/analysis: Uses GPT models to translate natural language into SQL or MongoDB queries, executes them, and returns results

- Visualization: Chart.js

- Forecasting: GPT models generate Python code, which runs securely inside the app

- Deployment: Docker or standalone bundle (includes app, Python, and local Postgres)

- Data support: SQL databases, MongoDB, JSON columns, Metabase dashboards

- Metabase integration: via Python wrapper for Metabase API (metabase_api_python)

Quickstart guide: https://databot.intellimenta.com/admin_guide/getting_started

GitHub Repo: https://github.com/Intellimenta/DataBot

Demo Video (4 min): https://www.loom.com/share/2271af6de3c04a2b8b07dbc48dfb2963

I'd love your feedback!