Settings
Manage integrations powering your knowledge base.
Supabase Storage
Private bucket for all uploaded documents.
pgvector + RAG
1536-dim embeddings, HNSW cosine index.
Gemini embeddings + chat
Lovable AI Gateway · embeddings + chat completion.
OCR (vision OCR)
Image + scanned PDF text extraction via Gemini vision.
Google Drive import
Pull documents from a Drive folder.
RAG configuration
Tuning for retrieval quality.
- Embedding model
- google/gemini-embedding-001
- Chat model
- google/gemini-3-flash-preview
- Vector dimensions
- 1536
- Top-K retrieval
- 6 chunks
- Chunk size
- 1000 chars · 150 overlap
- Index
- HNSW (cosine)