Founder-led product
SupportPilot — a support agent trained on your own documents
Upload your policies, price list and manuals; SupportPilot answers customer questions from them and shows exactly which page the answer came from. If it can't find it, it says so and brings in a person.
- Sector
- AI / Customer support
- Status
- Demo build
- Year
- 2026
- Scope
- AI design, RAG pipeline, Full-stack build
Built and owned by Xegency — not client work · shown with sample data
The problem
Support teams answer the same twenty questions all day, and a generic chatbot is worse than none — it invents refund policies that don't exist. Businesses need answers that are fast and provably true to their own rules.
What we built
- 01
Built a document library: PDFs, web pages and spreadsheets are read, split and indexed so the AI searches your material, not the internet.
- 02
Made every answer cite its source — document and page — so staff and customers can check it in one click.
- 03
Set a hard rule: if nothing in the documents answers the question, it says 'I'm not sure' and routes to a human instead of guessing.
- 04
Added a thumbs-up/down on each answer that feeds a list of gaps, telling you which document is missing or out of date.
Where it stands
The demo answers questions from a sample policy library with page-level citations and clean handover to a person. Looking for a first business with a real document set to pilot on.