Vespa.ai

What Vespa is

Vespa is a serving engine for search and retrieval rather than a search product. It combines vector search, keyword matching, ranking models and machine learning inference in one system, aimed at applications that must query large, constantly changing data with low latency.

What you can do with it

• Build hybrid search over big datasets
• Serve recommendations and personalisation
• Run ranking and inference alongside retrieval
• Power retrieval for generative systems
• Handle updates without full reindexing

Who it is for

• Enterprise search teams
• Recommendation and personalisation engineers
• Teams building retrieval backends

What to watch out for

• Vespa is infrastructure: deployment, tuning and on-call operations sit with your team
• Scale and latency claims need testing with your own data and query patterns
• Indexed content must respect the permissions of the source systems
• Ranking configurations are powerful and easy to get subtly wrong