Vespa AI

Vespa is an open-source engine for search, recommendation, and AI applications. Engineering teams use it to combine text search, vector (meaning-based) search, and structured data queries in one system, then apply machine-learned ranking to decide what to show first. It is built to handle very large data sets, even billions of items, and return results in well under a second, which makes it a common choice for retrieval-augmented generation (RAG) at scale. Yahoo originally built it in-house, and the independent company Vespa.ai now maintains it. Teams can download and run it themselves for free under an open-source license, or run it on Vespa Cloud, a managed hosting service on AWS.

This fits a company with real engineering capacity that needs to search or rank very large or fast-changing data sets beyond what an off-the-shelf search product handles well, and that has engineers comfortable running open-source infrastructure; it is a poor fit for a team that wants a point-and-click, no-code tool. A sensible starting point is running the open-source version on one well-scoped use case at no license cost, deciding later whether to move to managed Vespa Cloud. Vespa Cloud pricing was not published in enough detail for a planning figure at review time, so request current rates directly. The main risk is total cost of ownership: self-hosting has no license fee, but running it well takes real engineering time.