WeldCraft: Intelligent Document Search & AI Chat Assistant

Challenge

WeldCraft needed to make extensive technical documentation and web content instantly accessible. Keyword search missed contextually relevant results; users wanted conversational answers instead of sifting through PDFs.

Solution

We built an intelligent hybrid search system paired with a context‑aware RAG chatbot that:

• Hybrid Search Engine: Combined keyword (BM25) with semantic search to capture exact matches and conceptual relevance across PDFs and web pages.
• Conversational AI Interface: RAG‑based chatbot with conversation history that answers questions naturally using company documentation.
• Technology Stack: LangChain, OpenAI GPT‑3.5‑turbo, custom retrieval algorithms.

Results

• ✅ Users find relevant information 3× faster
• ✅ 85% reduction in documentation‑related support tickets
• ✅ Natural conversation interface replaces manual search across multiple documents
• ✅ Conversation memory for better multi‑turn UX

Technical Highlights

• Custom hybrid retrieval combining BM25 with vector embeddings
• Optimized chunking strategy for technical PDF content
• Conversation memory for multi‑turn dialogues
• Scalable architecture for concurrent users