Blog UzzAIBot / Flows
Knowledge Base — Train Your AI with Your Documents
How to upload PDFs and documents so the AI learns about your business and answers customer questions accurately.
05/03/2026•6 min•Por Equipe Uzz.Ai
Role para descobrir a jornada completa
SUMÁRIO6 seções
The UzzApp Knowledge Base uses RAG (Retrieval-Augmented Generation) technology to allow the AI to answer questions based on your company's own documents — product catalogues, policies, FAQs, procedures.
Supported formats
- PDF: product catalogues, manuals, policies.
- Text (.txt): procedures, scripts.
- Word (.docx): internal documents.
- Markdown (.md): technical documentation.
- Maximum file size: 50 MB per file.
How to add a document
- 1Go to Dashboard → Knowledge Base → "Upload Document".
- 2Select the file.
- 3Wait for processing (indexing can take 1–5 minutes).
- 4Test a question to verify the AI finds information correctly.
How vector semantic search works
When a customer asks a question, the AI transforms the question into a numeric vector and searches the knowledge base for the most semantically similar excerpts — even if the words used are different from those in the document.
Good practices for the knowledge base
- Keep documents up to date — outdated information will generate wrong answers.
- Organize content into clear, well-titled sections.
- Avoid extremely long documents — split by topic when possible.
- After uploading, always test with real customer questions.