AI-Assisted

Combine Oracle Machine Learning with embeddings to classify instantly for free, hand uncertain cases to a language model, and find topics.

Stop a language model's SQL from changing or revealing data with plan checks and a read-only schema, and measure how often it is right.

Let users ask questions in plain language by describing your schema to a language model and generating one SELECT statement in Oracle 26ai.

Check that RAG answers contain the right facts, cite real sources, and claim nothing unsupported, using SQL and a second language model.

Answer questions from your own data with Gemini and Oracle AI Database 26ai, citing sources and saying so when the answer is not there.

Get the details hidden in free text out with a language model in Oracle AI Database 26ai, in the right form and without invented answers.

Sort 400 tickets into categories, priorities, and sentiment with Gemini from Oracle SQL, in batches, and measure how far to trust the labels.

Log every language model call, retry when the provider is busy, see the tokens you pay for, and avoid calling a model for every row of a query.

Send prompts to a language model from SQL and PL/SQL in Oracle AI Database 26ai and control the role, randomness, length, and format of answers.

Find the passage that answers a question in PDF and Word documents with Oracle AI Database 26ai, and test which chunk size works best.

Turn PDF, Word, and HTML files into searchable chunks in Oracle AI Database 26ai with UTL_TO_TEXT, UTL_TO_CHUNKS, and VECTOR_CHUNKS.

Get the best of semantic and keyword search in Oracle AI Database 26ai with reciprocal rank fusion in SQL or a hybrid vector index.

Find exact codes, versions, and product names that semantic search misses, with an Oracle Text index, CONTAINS, and SCORE in Oracle Database.

Find answers by meaning in Oracle AI Database 26ai, learn how near is near enough, flag unanswerable questions, and keep filters accurate.

Get millisecond vector search with an in-memory HNSW index in Oracle AI Database 26ai, sized, measured, and tuned on 50,000 rows.

Speed up similarity search on large tables with an IVF vector index in Oracle AI Database 26ai, and measure what it gains and gives up.

Test embedding models on your own questions in Oracle AI Database 26ai and learn when a small in-database model is enough and when Gemini pays off.

Get Gemini embeddings from SQL and PL/SQL, embed whole tables in batches, call the REST API for extra options, and survive a busy provider.