heym

Why we included this project

Heym is a self-hosted, no-code workflow builder where AI is the execution core, not a bolt-on step. On a drag-and-drop canvas you wire together LLM calls, agent nodes, RAG pipelines, web scrapers, HTTP requests, and message queues into one runnable workflow, and you can model orchestrator-and-subagent patterns, run agent turns in parallel, or attach skills to agents in plain language. The operational side is where it earns its keep: every run leaves a trail you can inspect, with LLM traces, evals, logs, OpenTelemetry export, and cost tracking in real USD, so a workflow is something you can debug and audit rather than trust on faith. The same workflow can be triggered from the canvas, REST endpoints, SSE streaming, MCP clients, or a chat portal, and because it deploys on your own infrastructure, teams that want human oversight built into their automation get it without handing control to a hosted service.