DataBrief: AI-Powered Local Data Reports
Published:
Tech: Python · JavaScript · React · FastAPI · Ollama · LLM · local AI
DataBrief Tool
DataBrief is a privacy-first desktop application that analyses CSV files and generates clear, readable reports using a local AI model. It produces two types of output: a plain English summary for stakeholders, or a technical breakdown with code fixes for data teams — all without sending any data to the cloud.
The app runs entirely on your machine via Ollama, with a React frontend and Python/FastAPI backend. Uploaded files are read into memory and discarded immediately after the report is generated, so your data never leaves your computer.
Key features include:
- Dual report modes — stakeholder-friendly plain English summaries, or technical breakdowns with suggested code fixes
- 100% local AI — powered by Ollama (llama3), with no external API calls or subscriptions
- Privacy by design — files are never stored; nothing is sent to any external server
- Simple setup — one-command launch via
start.sh
My Contributions
- Designed and built a full-stack local AI application for automated data quality reporting • Integrated a locally-running LLM (via Ollama) with a React frontend and FastAPI backend • Prioritised privacy-first architecture — no cloud dependency, no data retention • Built for both technical and non-technical audiences with dual report modes.