Vespa.ai · The retrieval engine the RAG boom rediscovered — sixteen years of serving infrastructure now being asked whether it is also the right substrate for agents, which is a question worth watching someone answer in public.

One engine that does retrieval, ranking and inference in the same place. Most stacks bolt a vector database to a search index to a model server and pay for the hops; Vespa keeps the vectors, the text index, the structured fields and the ranking models — including ONNX models evaluated per document — inside the same nodes, so a query can score millions of candidates with a real model rather than re-ranking a shortlist. It is the engine spun out of Yahoo in 2023, where it had been running web-scale serving for years, and it is what Perplexity moved its search onto.

Sector

Memory & skills

Stage

Series A

On the roster since

15 Sept 2026