Dedalus Labs builds a compute substrate for AI agents, described as the fastest persistent computer for agents, enabling developers to host long-running, stateful AI agents with response times under 250ms. Its open-source SDK connects any LLM to MCP (Model Context Protocol) servers without Docker or complex configuration, letting developers build complex agents in roughly five lines of code with vendor-agnostic model handoffs and real-time streaming. Founded in 2025, Dedalus is based in San Francisco and was part of Y Combinator's Summer 2025 batch. The company raised an $11M seed round co-led by Kindred Ventures and Saga Ventures.