Skip to main content

​
SupaSherpa Vault RAG Feature Guide

​
Overview

SupaSherpa now answers your questions using your actual business documents - uploaded plans, briefs, financial models, and AI-generated sections - instead of generic advice. When you ask about revenue projections, pricing strategy, or target market, SupaSherpa pulls specific data from your vault and cites the source documents so you can verify every claim. Your documents stay private to your job, and SupaSherpa refreshes its knowledge automatically when you update files.

​
Step-by-Step Guide

  1. Upload documents to your vault - Navigate to your job’s data room or vault section and upload PDFs, Word docs, or other business documents. The system extracts text and processes them automatically within 2 minutes for most documents.
  2. Ask SupaSherpa a question - Go to the SupaSherpa page (/supasherpa) and type a question about your business, like “What is my projected revenue for year 2?” or “How does my pricing strategy align with my target market?”
  3. Review the document-backed answer - SupaSherpa searches your vault, retrieves relevant sections from your documents, and answers your question with inline citations showing which document and section the information came from.
  4. Verify the sources - Check the references section at the end of SupaSherpa’s response. Each citation includes the document name and section so you can validate the information against your original files.
  5. Update documents as needed - When you upload a new version of a document, the system re-processes it within 3 minutes. SupaSherpa automatically uses the updated content in future answers.

​
Common Questions

Q: What happens if I ask a question before uploading any documents?
A: SupaSherpa responds with general guidance and suggests uploading relevant documents to get more specific advice. You won’t see any document citations until you add files to your vault. Q: Can SupaSherpa access documents from my other jobs?
A: No. SupaSherpa only retrieves documents from the vault associated with your current job context. Your data stays isolated between projects. Q: What if SupaSherpa cites a document that doesn’t seem relevant?
A: SupaSherpa only includes citations when the similarity score exceeds 0.65. If a citation seems off, the system uses hedging language like “Your documents suggest…” instead of definitive statements. Upload more specific documents or rephrase your question for better results. Q: Does SupaSherpa use both uploaded documents and AI-generated content?
A: Yes. When you generate sections like executive summaries or market analysis through the platform, SupaSherpa searches both uploaded documents and generated content. Citations clearly distinguish between “Uploaded: Business Plan.pdf” and “Generated: Executive Summary.” Q: How long does it take for a new document to become searchable?
A: Most documents process within 2 minutes. A 20-page document typically completes chunking and embedding within 120 seconds. Large documents (50+ pages) may take up to 30 seconds for chunking alone.

​
Troubleshooting

Issue: SupaSherpa says my document failed to process
Check the document’s processing status in your vault. If text extraction failed (blank or corrupted PDF), re-upload a clean version. Image-only PDFs without OCR text will be marked as “SKIPPED” and excluded from search. Issue: SupaSherpa doesn’t find information I know exists in my documents
Try rephrasing your question with different keywords. The system searches based on semantic similarity, so “revenue forecast” and “projected income” might return different results. If you uploaded the document in the last 5 minutes, wait for cache refresh. Issue: Citations point to the wrong document section
This happens when similar content appears in multiple documents. Review the citation’s similarity score - scores between 0.65 and 0.75 indicate lower confidence. Upload more specific documents or consolidate related content into a single file. Issue: SupaSherpa response seems slow
First queries may take up to 500ms while the system generates embeddings. Identical follow-up questions use cached results and respond faster. If slowness persists, the system may be falling back to keyword search due to temporary API issues. Data Room & Vault Management - Organize your business documents, generate AI-powered sections, and track document versions. The vault feeds SupaSherpa’s knowledge base, so keeping your data room current improves answer quality. Brief Enhancement & Extraction - Upload source documents to extract structured briefs and generate enhanced business plans. These enhanced briefs become searchable vault content that SupaSherpa references when answering strategy questions. Charter Generation - Build your project charter using vault documents as source material. SupaSherpa can answer questions about your charter components and cite the underlying documents that informed each section.