YoMo
An open-source framework for building AI agents with LLM function calling, aimed at low-latency, geo-distributed serving of agent tools close to users.
YoMo is an open-source framework for building AI agents around LLM function calling, focused on speed, reliability and scalability. It lets a developer deploy and manage LLM tools and skills as serverless functions, and is designed to bring AI inference and tools closer to users through a geo-distributed architecture for faster response times.
Every request is encrypted with TLS 1.3, and the framework aims to streamline the lifecycle of LLM tools from development to deployment to reduce operational overhead. A CLI installs the framework, starts a local server that can use Ollama as the model provider, and scaffolds and runs a function-calling tool, for example a weather-lookup function that a chat request can then call.
It is written in Rust and built on QUIC for its underlying transport.