AI SaaS

DocuMind AI

A document intelligence platform with semantic search, Retrieval-Augmented Generation (RAG), and AI-assisted document understanding.

Next.jsLangChainOpenAIPineconeSupabaseVercel AI SDK

Kode Sumber Demo Langsung

Challenge

A legal team had 40,000 contracts in shared drives and no way to ask "which contracts auto-renew in the next 60 days and under what terms?"

Approach

I built a RAG pipeline: documents chunked and embedded into Pinecone, with Supabase storing source metadata. LangChain orchestrates retrieval and the OpenAI generator. A citation layer ties every generated sentence back to a chunk id.

Outcome

First-answer latency under three seconds on a 40k-document corpus.
Citations rebuilt on demand so answers are auditable.
Cut contract review prep time by ~70%.