Concepts of Oracle Simple Vector SDK (Oracle VecDB)

Understand the concepts behind Oracle AI Vector Search and how the Oracle Simple Vector SDK provides access to it.

Overview

Oracle VecDB is designed for vector-native use cases such as semantic search, retrieval augmented generation (RAG), knowledge assistants, and agent memory.

• Oracle AI Vector Search: architecture and access paths: Understand Oracle AI Vector Search and how it fits into the Oracle Simple Vector SDK.
• Embedding workflows: Choose integrated embeddings or bring your own vectors for Oracle VecDB.
• Embedding and reranking models: Understand the models used to generate vectors and optionally rerank search results.
• Vector Table: A vector table stores the records you want to search. Oracle VecDB supports integrated embedding vector tables and bring-your-own-vector tables.
• Indexes: Understand automatic indexing and choose between the default IVF and HNSW vector indexes.
• Record: A record is one row in a vector table. It represents one retrievable unit and consists of a record ID, a dense vector, and metadata.

Continue with the Python SDK

Start with the Quickstart: Build Vector Search with Python to install the SDK, configure the client, create a table, load records, and run a semantic search.

Then choose a task group:

• Configure the Python client
• Manage models
• Manage vector tables
• Manage data
• Search data
• Manage indexes

See the concepts in action

• Follow the Embedding & Ingest Patterns notebook to compare the integrated-embedding and BYO workflows.
• See the complete example applications and notebooks.