Knowledge Base (RAG)
Equip your agents with domain-specific knowledge using Retrieval Augmented Generation (RAG). Upload documents and let the AI answer questions instantly.
1. Ingest
Upload PDFs, DOCX, TXT files or crawl website URLs. We chunk and embed the text automatically.
2. Retrieve
When a user asks a question, we semantically search your vector database for relevant chunks.
3. Generate
The relevant context is injected into the LLM prompt, allowing it to generate an accurate answer.
Supported Formats
- PDF Documents: Product manuals, policy documents.
- Text Files (TXT, MD): FAQs, scripts, unstructured notes.
- Web URLs: Public documentation sites, help centers (sitemap scraping supported).
Embedding Model
We currently use Google's text-embedding-004 model for high-performance, multilingual vector embeddings. Storage is handled via ChromaDB.