For engineers
Heym Core
Build and run agentic systems.
The canvas and the runtime: agents, RAG, MCP, human review, traces and evals. Self-host it with Docker Compose or Kubernetes.
Free and source-available
Get startedSelf-hosted AI agent platform
Build agentic systems. Run them with confidence.
Sales / Leads and approvalsLeads and approvals
Heym is infrastructure for running and governing AI agents in production, with a visual editor for building them. It comes in two parts. Heym Core is the canvas and the runtime where engineers build and run agents. Heym Work is your team's AI workspace, the business interface where teams describe a job in plain words, get a tested workflow, work from pages of widgets built for them and shared with their team, approve what matters and keep a record of every run, all on top of Heym Core. Owners, Editors and Viewers see what their role allows, and a page can be shared with a person or a team with view or edit access. The window beside this text follows one job through both: a sales team asks in Heym Work Chat for a workflow that handles new leads, Heym Core builds and runs it with agents, and a reviewer approves the reply to the lead on a Heym Work page that holds a run form, a leads table and an approval card, where the run stays on record. Describe a workflow in natural language and the assistant builds it for you, or wire it yourself from a deep library of built-in nodes and integrations. Teams build on the same canvas together: everyone who opens a workflow shows as an avatar, edits reach every screen as they happen, each person's pointer moves in their own color, a click on an avatar follows that person's view, and the microphone lets people talk while they build. It supports LLM nodes for text generation and vision, Agent nodes with tool calling and Python tools, RAG for semantic search over Qdrant or built-in Postgres (pgvector), MCP client and server integration, human-in-the-loop approval checkpoints, content guardrails, parallel DAG execution, a skills system for portable agent capabilities, Playwright browser automation with auto-heal, and integrations with Telegram bots, Slack, IMAP inboxes, outbound WebSocket streams, email, Redis, RabbitMQ, Grist, and any HTTP API. Heym is source-available and fully self-hosted, so your data, agents, and runs stay on infrastructure you control, from the first prototype to production, with tracing and evals built in. Self-host with Docker Compose or Kubernetes on your own infrastructure.
Works with the models and tools you already use
Heym integrates with OpenAI, TypeSafe AI, Anthropic, Ollama, Cerebras, Google Gemini, OpenRouter, Slack, Discord, Gmail, Outlook, GitHub, Jira, Linear, Cloudback, Telegram, Notion, Qdrant, ClickHouse, Redis, Supabase, PostgreSQL, Amazon S3, RabbitMQ, WebSocket, Docker, MCP, Grist, Google Sheets, Google Drive, BigQuery, Playwright, Sentry, Ejentum.
The Heym platform
One platform, two ways in. Engineers build in Heym Core. Business teams work in Heym Work.
For business teams
Heym Work
Put agents to work, with people in control.
Teams describe a job and get a tested workflow, work from pages built for them, and approve what matters. It runs on your Heym Core.
Free trial available
See Heym WorkFor organizations
Enterprise
Custom AI systems for your business.
We design, build, and integrate AI-powered solutions around your workflows, with tailored interfaces, human oversight, and deployment support.
- Custom AI Solutions
- Integrations
- Production Pilots
- Ongoing Support
Custom engagement
Explore Enterprise SolutionsProduct tour
One campaign, end to end
One e-commerce sales campaign, followed end to end: workflow generation with the AI Assistant, human review, board execution, structured data, dashboards, traces, RAG, MCP, analytics, and team collaboration.
Heym Product Tour
Heym Core, for engineers
Build, run and debug agents on one canvas
Wire agents, tools, RAG and human review on a visual canvas, or describe the workflow and let the assistant build it. Every run is traced, step by step.
Try the interactive demoWatch the canvas execute branches, tools, memory, and HITL checkpoints as one graph. Debug execution traces live, with no abstraction layer in the way.
Live execution traceHealthy status
heym · visual-runtime-enginelive tracer
Describe a workflow and see Heym generate nodes, edges, and MCP tools on the canvas. Voice input and streaming responses are built in.
Prompt to canvasStreaming
heym · ai-assistantgenerating
Coordinate sub-agents, skills, sub-workflows, and human review from one orchestrator. Independent branches run in parallel.
Sub-agents and skillsParallel
heym · multi-agentorchestrating
The production basics stay in Heym Core, free and source-available: tracing, evals, approvals, single sign-on and audit logs. No sales call for the basics.
Source-availableSelf-hostable
heym · no-gatekeepingself-hosted
Heym Work, for business teams
Give every team a place to work with agents.
No canvas to learn. The platform team keeps sign-in, roles and the audit trail, on your own infrastructure.
- Ask in plain words. Heym builds the workflow and tests it before anyone relies on it.
- Approve what matters. A run pauses before an email goes out, a record changes or money moves.
- See every run. Each run keeps its steps, its duration and the instance that ran it.
Need more than an out-of-the-box experience? Our team can build tailored AI workflows, business interfaces, and integrations around your operations, powered by Heym Core.
Explore Custom AI SolutionsHeym Work/Lead intake
Example workflowChat
When a lead emails us, save it to the Leads table, draft a reply and ask me before it goes out.
- Saving the workflow
- Testing the workflow · attempt 1/5
- Finishing
Verified
Lead intake
Tested before it was saved.
Run by Heym Core, on your infrastructure
Source Available
Built for Your Control
Self-host Heym Core, review the code, and tailor the platform to your infrastructure.
Free to Use
Heym Core has no licensing fees and no usage limits. Self-host it on your own infrastructure with Docker Compose, a prebuilt container image, or Kubernetes.
Community Driven
Built together with contributors who share workflows, report bugs, suggest features, and shape the product roadmap through GitHub and Discord.
Source Available
Every line of Heym Core is open to read. Audit the security model, customize features, and contribute improvements back to the project.
Commons Clause
Heym Core is licensed under Commons Clause plus MIT: free to use, modify, and distribute, but not to resell as a paid service. Heym Work has its own commercial license.
$ git clone https://github.com/heymrun/heym.git
$ cd heym && cp .env.example .env
$ ./run.sh
✓ Ready at http://localhost:4017
Need every variable? See the environment reference.
Star on GitHub
Heym Core is licensed under Commons Clause + MIT
Knowledge Hub & Workspace
Build faster with ready to use templates and stay ahead with detailed articles from our engineering team.
Dev & IT Ops9 nodes
Agent-Guided Codex PR Dispatcher
Let an Agent inspect a GitHub request, prepare a confirmed Codex task, open a ready PR, and notify Slack when Codex needs input.
#Agent#Codex#GitHub
Marketing & SEO7 nodes
AI Viral Video Generator & TikTok Publisher
Read topics from Google Sheets, generate short videos with the VEO 3 API, and upload each one to TikTok automatically.
#TikTok#VEO 3#Video
Customer Support5 nodes
HITL Support Reply Agent
Draft a customer-facing support response, pause for human approval, then continue with the reviewed reply.
#HITL#Agent#Human Review
auto-ontologyontology
Auto Ontology: A Guide With Templates to Try
Learn what auto ontology is, how an AI agent draws the map of your business from calls and briefs, and import a sales and a marketing template to try.
November 7, 2026Read more
kubernetesai-agents
AI Agents on Kubernetes: 7 Questions for Your IT Team
AI agents on Kubernetes without the jargon. Learn what Kubernetes native means, why the Docker socket worries IT, and the 7 questions to ask any vendor.
November 3, 2026Read more
ai-liabilityai-accountability
Who Is Responsible When AI Makes a Mistake?
Who is responsible when AI makes a mistake? Six rulings and rules, 2,247 court decisions and five questions to answer first. Watch the 1-minute demo.
November 2, 2026Read more
Dev & IT Ops5 nodes
Adversarial PR Review
An orchestrator pulls a pull request from GitHub, a primary reviewer proposes findings, and a challenger on a different model disputes them — arbitrated into one calibrated approve, comment, or request-changes decision.
#Multi-Agent#Adversarial#Code Review
AI Agents5 nodes
Batch LLM Status Tracker
Send an array through the OpenAI Batch API, branch on live status updates, and collect the final per-item results.
#LLM#Batch API#Status Branch
Customer Support5 nodes
IMAP Support Inbox Triage
Watch a shared mailbox, summarize incoming support email, and route urgent messages to Slack.
#IMAP#Email#Slack
Learn and get help
Start fast. Ask when you get stuck.
Quick start
Run Heym Core on your machine and build a first workflow, step by step.
Read the guideTutorial videos
Short videos of real workflows, from document Q&A to human review.
Watch on YouTubeDiscord
Ask the Heym team and other builders, and share what you made.
Join DiscordGitHub
Read the code, report issues, and follow every release.
View on GitHub
Keep up with Heym
Product updates and practical agent ideas. One useful monthly email, no spam, and you can unsubscribe anytime.
Stop gluing AI tools together.
Start shipping workflows.
Build with Heym Core on your own infrastructure, or work with our team on a custom AI solution.
Docker Compose · Kubernetes · MIT + Commons Clause
Heym has two parts that run on your own infrastructure. Heym Core is where engineers build and run agentic systems, with agents, RAG, MCP, human review, traces, and evals. Heym Work is your team’s AI workspace, the interface business teams use to describe a job, get a tested workflow, work from pages of widgets built for them and shared with their team, approve what matters, and see a record of every run, and it runs on top of Heym Core.
Heym is source-available infrastructure for running and governing AI agents in production. You build agentic systems on a drag-and-drop canvas, or describe them to the AI assistant, by connecting a broad library of node types across seven categories (triggers, AI, logic, data, integrations, automation, and utilities) into production-grade workflows that run on your own infrastructure.
Every workflow is a directed acyclic graph (DAG) that Heym compiles, validates, and executes in parallel where dependencies allow, giving you maximum throughput without manual parallelism boilerplate.
Explore the platform overview, feature deep dives, node reference, documentation hub, use cases, enterprise options, and community resources from the links in the site footer.
MCP integration works in both directions. As an MCP client, Agent nodes connect to any external MCP server and gain access to all tools it exposes. Heym automatically discovers available tools on connection and presents them in the agent configuration panel.
As an MCP server, Heym exposes all published workflows as callable tools over the MCP protocol. Claude Desktop, Cursor, VS Code extensions, and other MCP-compatible clients can invoke your workflows directly from their native interfaces. The heym_chat tool brings Heym’s Chat into those clients as well: from one message it can build and test a workflow, or create, edit, and run prompt and agent eval suites, and it returns the Evals link of every suite it saves. With Enable A2A on, the same Heym Agent answers other agents over A2A 1.0, the Agent2Agent protocol, with an agent card for each MCP server and task progress streamed as it works.
The skills system enables portable, reusable capability bundles for Agent nodes. A skill is a zip archive or Markdown file containing a SKILL.md instruction document and optional Python tool files. Dragging a skill onto an Agent node extends that agent’s system context and toolbox without modifying the node configuration.
Skills can define custom tool schemas, persistent memory structures, and multi-step reasoning templates. Agent nodes include an AI Skill Builder modal for drafting and revising skills with live previews. The same skill can be shared across multiple agents and workflows for consistent behavior across an organization.
Join the Heym Discord community to share workflows, report issues, and collaborate with other builders. Tutorial videos walk through canvas basics, agent configuration, RAG setup, and deployment patterns.
Community templates and blog posts demonstrate real production patterns — from MCP tool servers to evaluation-driven prompt iteration. Sharing workflows publicly helps other teams adopt guardrails, tracing, and human review patterns that are easy to skip in early prototypes.
Contributing back to the source-available repository improves documentation, adds node types, and hardens executor edge cases. Whether you file a bug with a minimal reproduction or ship a pull request for a new integration node, the community benefits from shared execution semantics tested across diverse deployments.