FeaturesDocuments & RAG
Documents & RAG
Document Management
Upload documents to your ThinkFleet workspace for retrieval-augmented generation (RAG). Your agent can search and reference uploaded documents when answering questions.
How It Works
- Upload — Upload documents through the dashboard (PDF, text, markdown, etc.)
- Processing — Documents are automatically chunked and embedded as vectors
- Storage — Chunks are stored in a vector database (PostgreSQL with pgvector)
- Retrieval — When your agent needs information, it searches relevant document chunks
- Generation — The agent uses retrieved context to generate informed responses
Document Categories
| Category | Use Case |
|---|---|
| Reference | Knowledge base articles, manuals, guides |
| Training | Examples for the agent to learn from |
| Writing Style | Samples that define tone and voice |
Storage Limits
| Plan | Document Storage |
|---|---|
| Free | Not available |
| Pro | 5 GB |
| Lifetime | 20 GB |
| Enterprise | 100 GB |
Document Status
Documents go through a processing pipeline:
- Pending — Uploaded, waiting for processing
- Processing — Being chunked and embedded
- Ready — Available for search and retrieval
- Error — Processing failed (check document format)