Set up a RAG pipeline

What is RAG?

RAG combines model generation with document retrieval so answers are grounded in your own data.

1) Create knowledge base

1. Dashboard → Knowledge Bases → New
2. Set name and description
3. Save

2) Upload documents

Supported: PDF, TXT, Markdown, DOCX.

After upload, Mycelis:

1. chunks content
2. creates embeddings
3. stores vectors in Qdrant

3) Enable KB on agent

1. Open your agent
2. Go to Knowledge & Tools
3. Enable KB
4. Save

Best practices

• keep chunks short for precision
• use structured documents
• add prompt instruction to prefer provided context

Next steps

• Configure first agent
• MCP tools guide