Datasnap

DataSnap turns a business question and a tabular dataset (CSV, XLSX, Parquet, TSV, JSON, or JSONL) into a decision-ready brief. It profiles the data, plans the analysis, generates Polars/SciPy/Matplotlib code, and runs that code in an isolated, timed subprocess that self-corrects on runtime failures. A LangGraph workflow orchestrates the steps, using Gemini with an automatic fallback chain for transient provider errors. The result is a verified report with an executive summary, direct answer, statistical insights, charts, metric tables, and recommendations — downloadable as Markdown or HTML and savable to a local browser library. It ships as a FastAPI backend with Redis-backed anonymous quotas and a dark-first Vite/React frontend, requiring no user account or database.