Vector versus hybrid search indexes
Key Takeaway
A vector search index matches by meaning and suits long or general queries. A hybrid search index adds keyword search, so choose it when queries also contain exact terms like product codes.
Data 360 builds search indexes over your company's data so that AI tools like Agentforce agents and Prompt Builder can find relevant content to ground their responses. A vector search index matches content by meaning. A hybrid search index combines vector search with keyword search, so it can also match exact terms like product codes. You choose between them based on your data and the kind of queries users send.
What Data 360 search indexes do
Data 360 search indexes ground generative AI in your own customer-specific data. The data can include unstructured data, such as text documents, chat logs, web content, and customer feedback. You can create a vector search index or a hybrid search index, depending on your data and query needs.
After either index is built, apps such as Prompt Builder, AI agents, and Tableau can search that data. Data 360 search is fundamental to Retrieval Augmented Generation (RAG).
Common uses include the following.
- Service agents search knowledge articles, emails, support tickets, and chat logs for precise answers.
- Sales agents query emails, call transcripts, and notes to build meeting briefings.
- Automation flows find semantically similar Cases to flag duplicates and classify records.