Your customers and your team ask the same questions all day, and the answers are already sitting in your PDFs, help centre, policies or website. I build an assistant that reads those documents and answers in plain language — with the source cited every time, and an honest "I don't know" when the answer isn't there, instead of a confident guess.

What you get: • Your documents, FAQs and website ingested into a private knowledge base, re-indexed when they change • A chat assistant deployed where you need it: a widget on your website, Slack, or WhatsApp • Every answer linked to the page or document it came from • Off-topic and unanswerable questions handled safely, with handover to a human if you want it • A retrieval evaluation report before launch: a set of real questions, and the measured rate at which the assistant finds the right source — so you know it works before your customers do • Admin view of conversations and unanswered questions, so the knowledge base improves over time • Source code, deployment, and a handover document

Built with LangGraph, a vector database (Qdrant or pgvector), and the LLM of your choice — OpenAI, Claude, or an open-weight model on your own server if your data can't leave it.

Before we agree a price, I'll build a working demo on your actual content so you can try it. The starting price covers up to 50 documents or pages and a web widget; larger document sets, extra channels, custom branding, and self-hosting are quoted after the demo.

You can try the same system right now on my own site: anshmangukiya.in — ask it anything about my work.