We're Burak and Ceren two engineers based in Berlin. We built Heym because we kept running into the same wall.
Every time we needed agents, document retrieval, approval steps, and observability all in the same workflow, we ended up gluing together four or five different tools. Existing platforms were built for simple, rule-based logic. The moment a workflow became AI-native, we were fighting the platform instead of building.
So we built what we actually wanted to use.
Heym is a self-hosted AI workflow automation platform. Visual canvas, multi-agent orchestration, built-in knowledge retrieval, human review checkpoints, full execution traces, and the ability to expose any workflow as a tool your AI assistant can call — all in one place, running on your own infrastructure.
Every decision reflects a constraint we hit in practice. Review checkpoints exist because some decisions need a human sign-off before they proceed. Automatic context management exists because we watched long-running agents silently fail mid-task. The evaluation system exists because we got tired of manually checking whether a prompt change broke something downstream.
This is v0.0.1 — actively developed, source-available.
If you're building AI workflows and spending more time on glue code than on the actual problem this is for you.
Happy to answer questions about the architecture or any design decisions. 🙏
— Burak & Ceren
Comments (1)
Visual canvas for AI workflow automations is a smart approach. Makes complex agent pipelines much easier to reason about without needing to write code for every connection.
@chaudharyarun5797 Indeed it is and thank you :)