Document Chat System (document-chat-system.vercel.app)

Document Chat System is a production-ready, open-source platform for uploading documents, running semantic vector search, and chatting with AI about their contents. Built with Next.js 15, React 19 and TypeScript (end-to-end TS + Zod validation), it automates document processing — extracting text/metadata, chunking, embedding, and indexing — so users can drag-and-drop PDFs, Word files, images and more, then query content in natural language with answers citing source documents. The stack emphasizes performance and scale (Upstash Redis for low-latency responses, CDN distribution, edge functions, live doc sync and chat streaming) and ships with a production-optimized Dockerfile and Vercel deployment scripts. Significant for AI/ML teams and builders: it’s fully open-source (MIT), self-hostable to avoid vendor lock-in, and model-agnostic — swap between OpenRouter (100+ models), OpenAI, Anthropic or ImageRouter seamlessly. Multi-tenancy, vector search, automated pipelines, and optional Stripe-based billing (Prisma-powered pricing plans and subscription controls) make it suitable both as an internal knowledge-management/customer-support tool and as a turnkey SaaS product. The project’s combination of real-time semantic retrieval, streaming chat, and monetization hooks lowers the barrier for deploying custom, privacy-controlled document AI services.