heym

Thanks for taking the time to review Heym in such detail. We really appreciate the thoughtful breakdown, especially the points around AI-native orchestration, runtime inspection, self-hosting, and production workflows. There are a few parts we would love to clarify or update, as the product has evolved quite quickly: • Heym is no longer positioned primarily as a workflow canvas. The canvas is one interface, but the broader product is a self-hosted runtime for building, running, observing, evaluating, and governing agentic systems. • The integration surface is now much broader than an AI-focused set. Heym supports integrations such as Slack, Gmail, Outlook, GitHub, Jira, Linear, Notion, Google Drive, PostgreSQL, S3, Redis, RabbitMQ, Playwright, and more, alongside MCP, skills, and custom nodes. • Self-hosting does not require teams to operate a queue or object storage layer by default. For example, multi-instance deployments can use PostgreSQL without requiring an external broker, while file storage can also use local volumes. • The section around pricing data becoming stale may also be worth revisiting. Model pricing is maintained through a synced pricing table and is used for real-time cost calculation. • We completely understand mentioning the pre-1.0 status as something teams should consider. We would just suggest framing API or schema instability as something to verify rather than as an existing limitation, since that conclusion is not necessarily implied by the current version number alone. Overall, thank you again for covering Heym. The article captures several important parts of what we are building, and we would be happy to help with any technical details if you would like to update these sections.