Drop
Clean data. Clear insights. Immediate action.
AI-powered data analysis
Drop
Interactive preview — try the live demo for the full experience
Drop is a self-serve data analysis platform that turns raw spreadsheets into actionable insights in minutes. Users drag-and-drop CSV, XLSX, or JSON files and receive instant statistical profiling — schema detection, missing values, correlations, outlier analysis, and a letter-grade data quality score. AI generates executive summaries, prioritized insights, ML use-case recommendations, and auto-built charts on a draggable dashboard grid. A floating chat assistant lets users ask natural-language questions about their dataset. Anonymous trials expire in 24 hours; authenticated users get persistent history, sharing, favorites, and a credit-based quota system across Free, Paid, and Enterprise tiers.
Zero-to-insight upload flow with sessionStorage handoff for instant results-page rendering
AI Discovery suite — executive summaries, prioritized insights, and ML recommendations via async Cloud Tasks
Smart chart dashboard — drag-and-resize grid with Recharts and layout persistence
Split architecture — independent Next.js frontend and Django API repos, each with Cloud Build → Cloud Run pipeline
- ›Drag-and-drop upload for CSV, XLSX, and JSON with instant statistical profiling
- ›Data quality scoring with S–F grades and component breakdown (completeness, uniqueness, consistency)
- ›AI insights — executive summaries, ML recommendations, and prioritized findings with quality alerts
- ›Auto-generated smart charts (bar, line, pie, scatter) on a responsive dashboard grid
- ›Conversational data exploration — floating chat with suggested questions and multi-turn history
- ›Multi-channel auth — email/password, Google, GitHub, magic link, and QR-code mobile login
- ›Saved analyses with search, favorites, public/private toggle, and shareable links
- ›Freemium credit model — Anonymous, Free, Paid, and Enterprise tiers with usage tracking
- ›Five curated sample datasets for exploration without uploading files
- ›Async task polling (HTTP 202 + Cloud Tasks) with exponential backoff for long-running AI jobs
The Next.js frontend uses a typed Axios service layer; upload responses are cached in sessionStorage so the results page renders immediately. `NEXT_PUBLIC_API_URL` is baked in at Docker build time for standalone Cloud Run deployment. The Django REST API runs pandas-based profiling synchronously on upload, then offloads AI work to Cloud Tasks workers (`ai_insights`, `ai_chat`, `chart_generation`). Clients poll `/api/tasks/{id}/` until completion. Files live in Google Cloud Storage with signed URLs; auth uses JWT with refresh rotation across five login channels.
- 01SessionStorage as an upload cache eliminates the results-page loading flash and handles React Strict Mode cleanly
- 02Async AI via Cloud Tasks (202 + poll) keeps the UI responsive and avoids Cloud Run request timeouts
- 03Build-time `NEXT_PUBLIC_*` injection is required for Next.js standalone output on Cloud Run — runtime env alone is not enough