DBDraft — pending technical review

Oracle 26ai Vector Search for app developers

Build RAG and semantic search right inside the database with native VECTOR columns and similarity queries.

Intermediate⏱ 1 min readUpdated: 2026-06-09

Oracle 26ai adds a native VECTOR type and similarity search. You can embed text, store the vectors beside your rows, and run semantic queries in plain SQL — no separate vector store needed.

Add a VECTOR column

ALTER TABLE docs ADD (embedding VECTOR(1536, FLOAT32));

Populate and query

Generate embeddings (via a model) for each row, store them, then find the nearest neighbours to a query vector with VECTOR_DISTANCE.

SELECT id, body
FROM   docs
ORDER  BY VECTOR_DISTANCE(embedding, :query_vec, COSINE)
FETCH FIRST 5 ROWS ONLY;

💡 Keep it grounded

Return the matched rows AND cite them to the user — that's what makes a trustworthy RAG answer (it's exactly how Ask AI on this site works).

Check your understanding

Check your understanding

0% · 0/2

Where do the vectors live in 26ai?

Which function finds nearest neighbours?

Need this delivered?

Request a quote
← Upgrade an APEX 24.1 app to APEX 26.1CI/CD for APEX with GitHub Actions →