Inspiration
While working on a research project, I had thousands of lines of raw data to analyze manually. I realized there should be a simpler way to query and visualize data—like Cursor, but built for databases.
What it does
Fathom is an AI-powered desktop app that runs locally. Users can upload a database, view their tables, generate charts, and ask natural language questions. The AI agent analyzes the database and delivers answers instantly—no SQL required.
How we built it
We built Fathom with a React frontend and a Python backend. The backend integrates with the OpenAI API for natural language understanding, while the frontend provides an interactive desktop interface. Everything is packaged for cross-platform use.
Challenges we ran into
- Mapping natural language queries to complex database structures.
- Ensuring performance while handling large datasets.
- Building a smooth local-first experience with no data leaving the user’s machine.
Accomplishments that we're proud of
- A functional cross-platform desktop app that feels seamless to use.
- Successfully combining charts, AI-powered Q&A, and secure local execution.
- Creating a tool that can save researchers, students, and developers hours of manual work.
What we learned
We learned how challenging it can be to bridge the gap between human language and structured database queries. We also gained experience with building cross-platform desktop apps and ensuring privacy by keeping everything local.
What's next for Fathom — Your AI-powered database assistant
- Expanding support for more database types beyond SQLite.
- Adding advanced visualization options and export features.
- Improving the AI agent to handle multi-step, context-aware queries.
- Building collaboration features so teams can analyze data together.
Built With
- electron
- fastapi
- openai
- plotly
- python
- react
- sqlalchemy
- tailwindcss
- typescript
Log in or sign up for Devpost to join the conversation.