Loading DocLumi...
Loading DocLumi...
We believe everyone deserves to understand every document they encounter — regardless of language or complexity.
DocLumi was built in 2026 by an independent developer who grew frustrated with existing PDF tools. Most tools either charge too much, lack AI features, or don’t work well with non-English documents. DocLumi combines the best of all worlds: powerful AI, bilingual translation, and citation tracking — at a fraction of the cost.
Built with Next.js 16, Supabase PostgreSQL, DeepSeek V4 (1M token context), and deployed on Vercel’s global edge network. Open source on GitHub. GitHub.
Tech Stack: Next.js 16 (App Router) · DeepSeek V4 (MoE, 236B params) · Supabase (PostgreSQL + RLS) · Vercel (Edge Network) · TypeScript · Tailwind CSS · PyMuPDF. View on GitHub.
Sources: Chatbot Arena (lmsys.org, 2026.06) · Princeton KDD 2024 · Tokyo GEO-SFE 2026 · Ahrefs 17M Page Analysis · WMT25 Shared Task