Heym
Build agentic systems. Run them with confidence
145 followers
Build agentic systems. Run them with confidence
145 followers
Heym is a source-available platform for building and running agentic systems on your own infrastructure. Build multi-agent workflows visually, connect your data and tools, bring in coding agents like Codex and OpenCode, and add human approval where it matters. See exactly how every run performs with built-in traces, costs, latency, and evals. Self-host Heym, use your own models and credentials, and ship workflows as portals, APIs, or MCP tools.
This is the 2nd launch from Heym. View more
Heym
Launching today
Heym is a source-available platform for building and running agentic systems on your own infrastructure. Build multi-agent workflows visually, connect your data and tools, bring in coding agents like Codex and OpenCode, and add human approval where it matters. See exactly how every run performs with built-in traces, costs, latency, and evals. Self-host Heym, use your own models and credentials, and ship workflows as portals, APIs, or MCP tools.
Free
Launch Team / Built With
Heym
Hi Product Hunt 👋
We built Heym because creating an AI workflow is only the beginning. The hard part is connecting agents, company data, tools, approvals, and production infrastructure without ending up with a stack of disconnected products.
Heym brings these pieces together in one source-available platform that runs on your own infrastructure. Individual users can build and operate complete agentic workflows, while teams can collaborate through shared workflows, human approval steps, and an Agentic Kanban Board that coordinates work between people and agents.
You can design multi-agent workflows visually, connect RAG, MCP tools, APIs, and coding agents, then inspect every execution through traces, token usage, model costs, latency, and evals.
For this relaunch, we added first-class Codex and OpenCode nodes, an Agentic Kanban Board that runs workflows as work moves between stages, bidirectional MCP support, live production run inspection, and improved Docker and Kubernetes deployment.
Heym is free to self-host. You can use your own models, credentials, and infrastructure, and expose the same workflow through portals, APIs, or MCP.
We would love to hear what you are building, which integrations you need next, and what would make Heym more useful for you or your team.
@ceren_kaya_akgun nice launch congrats🙌Self-hosting via Kubernetes is a major plus, do you provide official Helm charts or operator definitions for scaling background workers dynamically?
Heym
@priya_kushwaha1 Thanks, Priya. Really appreciate it.
Kubernetes is an important part of the deployment path for teams running Heym at larger scale. We support self-hosted Kubernetes deployments and can help shape the setup around the workload, including multi-worker topology, resource planning, and autoscaling.
We keep the core product self-hosted and available rather than placing those capabilities behind an enterprise tier. Where enterprise support adds value is in the hands-on deployment work: fitting Heym into an existing cluster, security and networking model, observability stack, and scaling policy.
Since every environment is a little different, we approach Helm and operator configuration as part of the deployment architecture instead of treating a generic package as the complete answer. Happy to compare notes on your setup and point you to the best starting path: github.com/heymrun/heym
@ceren_kaya_akgun Thanks for the detailed explanation, that sounds like a really flexible approach for teams managing larger Kubernetes deployments. all the best for launch🙌
Heym
Hi everyone 🤗 One thing worth highlighting about Heym is where we are headed.
More capable models like DeepSeek V4 Flash 0731, combined with two DGX Spark systems, make truly long running local AI inference workflows possible. Coding agents can continuously build, test, improve, and ship software. General purpose agents can handle content, research, and other recurring operational work.
We believe AI automation should remain open, accessible, self hosted, and privacy minded.
Heym does not collect telemetry, and we never will. More importantly, we do not hide essential capabilities behind enterprise gatekeeping or closed sales calls: https://github.com/heymrun/heym#no-enterprise-gatekeeping
Thank you for your support, upvotes, and belief in us. Please keep supporting Heym as we build what comes next.
Heym
Hi Product Hunt! 👋
Thanks for checking out Heym. We built it for teams that want to turn AI ideas into reliable visual workflows, with agents, RAG, MCP tools, human approvals, traces, and self-hosting in one place.
If you’re comparing Heym with n8n, this overview is a good place to start:
https://heym.run/compare/n8n
Here are our three most-watched tutorials:
• Agentic AI Workflow Automation with Heym Boards
• Heym Dashboards: Build Interactive Dashboards with AI
• Heym Analytics: Monitor and Improve Your AI Workflows
We’d love to hear what you’d automate with Heym, and we’re happy to help you get started.
Congrats on the launch! The combination of self-hosting, human approvals, and built-in traces/evals feels especially strong for teams that want agentic workflows without losing control.
Being able to ship the same workflow as a portal, API, or MCP tool also makes Heym feel much more production-oriented than a typical visual agent builder.
Heym
Thank you @alpertayfurr, this is exactly the direction we’re aiming for with Heym. We want the visual builder to be the entry point, but the bigger goal is to give teams one place to orchestrate agents, tools, data, and human decisions while keeping every execution inspectable and controllable. (GitHub)
A good example is our Adversarial PR Review template, where an orchestrator coordinates multiple reviewer agents through GitHub MCP and the full execution can be inspected through traces:
https://heym.run/templates/adversarial-pr-review
Really appreciate you calling out the portal, API, and MCP side too. Being able to build once and expose the same system through different interfaces is a big part of how we think about production use.
Cloudback
Congrats on the launch! Lovely to see products like this being open-sourced.
Heym
@elv1s42 Thank you so much, Evgeniy! It truly means a lot. We believe deeply in building Heym openly with the community, and we really appreciate your support.
minimalist phone: reduce your screentime
Wishing you good luck with the launch.
Heym
@busmark_w_nika Thanks a lot! We really appreciate your support.