Using RAG

Guide · Developers and operators · Applies to 5.0.x

RAG is the standard way to make Eldric answer from your own documents instead of relying only on model training.

Upload documents

Create or select a knowledge base, upload supported documents and wait until indexing finishes. For larger uploads, use the public upload and vector ingest endpoints.

curl -X POST https://<your-eldric-host>/api/v1/vector/ingest \
  -H "X-API-Key: $ELDRIC_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"tenant":"acme","namespace":"policies","text":"The text you want indexed...","source":"uploaded-document"}'

Ask grounded questions

Select the knowledge base in chat or call the API with the intended namespace. When retrieval finds relevant passages, the response should include citation context.

Troubleshooting

1. Confirm indexing is complete.
2. Confirm the caller has access to the tenant and knowledge base.
3. Try language that appears in the source document.
4. Check known issues for current retrieval limitations.