Introducing Teel – Turbo-Efficient Execution Library for Agents (but not only)

Teel is an open-source Python runtime that accelerates AI agent execution by automatically overlapping independent, I/O-bound calls. It allows developers to write plain sequential code without `async` or `await` by using decorators to mark expensive functions and orchestration logic. A shadow interpreter reads ahead in the Python bytecode, identifies calls with known arguments, and dispatches them on background threads before the main execution reaches them. This approach significantly reduces wall-clock time for tasks with fan-out, fan-in patterns, common in agent swarms and complex reasoning frameworks. In benchmarks, Teel demonstrated up to an 8x speedup on a Tree of Thoughts implementation and proved faster than manually written asynchronous code in a LangChain Deep Agents example.