Vespa

Vespa is a distributed serving engine for AI applications. It combines database, search, and machine-learning inference in one platform, so teams can build applications that retrieve, rank, and serve data in real time. The engine handles text, vectors, tensors, and structured data, and it scales to billions of data items while keeping query latencies below 100 milliseconds. It is the technology behind an AI search platform that powers use cases like search, RAG, recommendations, and personalization.

Under the hood, Vespa unifies retrieval and ranking with integrated machine-learned model inference. That means you can run relevance models directly where the data lives, rather than moving data between separate systems. It supports hybrid search, multi-vector representations, and tensor operations, which makes it useful for modern generative AI pipelines that need more than simple vector similarity. Vespa is available as open-source software on GitHub, and the company also offers a managed cloud service, including on AWS.

Developers and data platform teams use Vespa when they need search or recommendation systems at serious scale. It is a good fit for building RAG applications, powering AI agents, serving personalized content, or replacing a patchwork of search and vector databases with one engine. The free trial and documentation make it easy to start small and grow.