A first production RAG assistant shows why retrieval design matters
A customer-facing is being built for sustainable and natural building materials. It needs to answer technical building questions about installation methods, material , construction details, thermal , moisture management, and similar topics. The contains about 30 years of specialist experience.
The source material is mixed and difficult to handle. It includes two scraped websites with articles, guides, FAQs, and case studies, adding up to several hundred pages of structured written content. It also includes hundreds of product technical data sheets in PDF form, often with tables for thermal values, ive strength, lambda values, and vapor resistance.
Detailed installation guides and method statements add step-by-step procedural material. The weak point appears to be and chunking, not just the final answer writing.
Key points
- The assistant must answer customer questions about technical building materials and installation details.
- The represents about 30 years of specialist knowledge.
- The content includes scraped websites, hundreds of PDFs, and detailed installation procedures.
- Table-heavy PDFs and step-by-step guides make and chunking harder than plain text.
- Better can reduce wasted tokens by sending only the most relevant evidence to the model.