Hornet is a retrieval engine designed for AI agents and the workloads they create. Unlike traditional search infrastructure, which is optimized for short human queries and fast, top-ten results, Hornet supports agents that issue long, structured queries, retrieve complete documents, and run iterative or parallel search loops. Its schema-first APIs are intended to make integrations more predictable and reduce errors and wasted tokens.

The platform offers retrieval methods for a range of uses, from single-agent reasoning over focused datasets to multi-agent systems handling high query volumes across web-scale data. Hornet can be deployed as a managed service, on-premises, or in a customer’s own cloud, and is designed to run alongside agents and data. It works with models of different sizes.

Hornet draws on its team’s experience building and scaling search technologies for large consumer products and search engines. The company is developing its platform for the changing needs of agent-based search and invites interested users to subscribe for product updates and a behind-the-scenes look.