AI-assisted workflow generation
Describe a workflow in natural language or voice and the assistant generates nodes and edges on the canvas. The assistant can stream its output and apply valid workflow JSON directly to the editor.
Heym
Heym is a source-available AI workflow automation platform for building visual, inspectable automations with agents, RAG, and approval checkpoints. It is designed for teams that want to self-host workflow infrastructure on their own systems.
Overview
Heym is an AI workflow automation platform for building and running agentic systems on a visual canvas. It combines a low-code editor with an AI assistant that can generate workflows from plain language, plus manual editing for teams that want direct control over nodes, edges, and execution paths.
The platform is positioned for AI, platform, and automation teams that need inspectable execution, human approval checkpoints, multi-agent orchestration, built-in RAG, MCP support, and source-available self-hosting. The product pages also describe tracing, evals, dashboards, memory, guardrails, and deployment on infrastructure you control.
Features
Visual workflow editor
Build or inspect workflows on a drag-and-drop canvas powered by Vue Flow. The editor includes built-in node types, expression-based data transforms, version history, data pinning, and extraction of selections into reusable sub-workflows.
Multi-agent orchestration
Coordinate one orchestrator with named sub-agents and sub-workflows. Agents can call tools, use Python tools, connect to MCP servers, load skills, and hand off work across multiple levels of nesting.
Built-in RAG pipeline
Run semantic search over managed vector stores using Qdrant or built-in Postgres with pgvector. The RAG pipeline accepts PDF, TXT, Markdown, CSV, and JSON uploads and can use metadata filters and optional Cohere reranking.
Human-in-the-loop controls and memory
Pause workflows for review, generate public approval links, and let reviewers accept, edit, or refuse before execution continues. The platform also includes content guardrails and graph-based persistent memory for agents.
Deployment, observability, and integrations
Deploy workflows as public chat portals, monitor executions with traces and dashboards, and connect to supported services such as Slack, Gmail, GitHub, Jira, Telegram, Google Sheets, Playwright, PostgreSQL, Redis, RabbitMQ, S3, BigQuery, and MCP endpoints.
Use cases
Design workflows from natural language
Build an internal automation that starts from a plain-language prompt, then refine the generated graph by hand on the canvas. This suits teams that want AI to speed up workflow creation without giving up visual control.
Orchestrate multi-step agent operations
Create long-running agent systems that delegate work to sub-agents, call tools, and pause for review before critical actions. This is useful when tasks need both autonomy and checkpoints.
Add retrieval to knowledge-driven workflows
Connect a workflow to company knowledge stored in Qdrant or Postgres pgvector, then use RAG nodes to retrieve relevant context for LLM or agent steps. This fits search, support, and knowledge-assist flows.
Insert human approval into production flows
Run workflows that require human review before anything is sent, approved, or executed. The platform exposes approval checkpoints and public review links, which is useful for support replies, compliance-sensitive tasks, and content approval.
Publish and connect workflows to existing tools
Expose a workflow as a public chat portal or use integrations such as Slack, Gmail, GitHub, Jira, Sheets, Playwright, or APIs to connect it to existing systems. This is suited to teams operationalizing AI across multiple entry points.
Pros and Cons
Pros
- Combines visual workflow building with an AI assistant that can generate nodes from natural language.
- Supports multi-agent orchestration, RAG, MCP, human approval checkpoints, and persistent agent memory in one platform.
- Can be self-hosted on your own infrastructure with Docker Compose or Kubernetes.
- Includes tracing, dashboards, and eval-oriented workflow visibility for inspection and debugging.
- Offers a documented library of nodes and guides, with the docs page showing broad coverage across workflow and platform topics.
Cons
- The collected pricing URL returns a 404 page, so the available sources do not show pricing plans or commercial packaging.
- Some areas are only partially documented in the collected sources, including integrations details, deployment variants, and limitations beyond self-hosting.
FAQ
How do you build workflows in Heym?
Heym is designed for building AI workflows visually. The source material describes a low-code canvas, an AI assistant that can generate nodes from plain language, and a library of built-in nodes and integrations for wiring workflows by hand when needed.
What kinds of workflow components does Heym support?
The documentation and product pages show workflows built from triggers, AI nodes, logic, data steps, integrations, automation, and utility nodes. Heym also supports multi-agent orchestration, RAG, MCP client and server connections, human approval checkpoints, and parallel execution.
Can Heym be self-hosted?
Yes. The site states that Heym is source-available and fully self-hosted. It can be deployed with Docker Compose or Kubernetes on infrastructure you control, and the enterprise page emphasizes deployment on your own infrastructure with your own models and credentials.
Does Heym support team collaboration and approvals?
The product pages describe support for team collaboration through shared vector stores, dashboards, workflow traces, human review checkpoints, and public review links for approvals. The source does not provide detailed role or permission documentation on the pages reviewed.
What does Heym cost?
A pricing page was not available from the collected sources; the link returned a 404 page. The homepage does state that Heym is free to use when self-hosted and licensed under Commons Clause plus MIT, but no hosted pricing details were shown.
Quick Facts
- Category
- AI workflow automation platform
- Product type
- Source-available, self-hosted software
- Primary users
- AI, platform, and automation teams
- Deployment
- Docker Compose or Kubernetes on your infrastructure
- Core capabilities
- Visual workflows, multi-agent orchestration, RAG, MCP, HITL approvals
- Website
- heym.run
Alternativas a Heym
Codex Plugins
Codex Plugins bundle reusable skills, app integrations, and MCP servers into workflows you can install in the Codex app or use from Codex CLI. They help extend Codex with connected-service tasks, reusable instructions, and shared team workflows.
Awish
Awish is an AI-powered no-code automation platform that turns natural-language requests into automated workflows across connected apps. It serves individuals and teams that want to automate repetitive tasks, chat-based assistants, and workflow updates without manual configuration.
CreateOS Sandbox
CreateOS Sandbox is an isolated compute environment for running code and agent workloads inside Firecracker micro-VMs. It is designed for workflows that need machine-level isolation, private networking between sandboxes, and programmatic control through SDK, CLI, or MCP.
Struere
Struere is an AI-native platform for turning spreadsheet data into structured operational software with dashboards, alerts, and automations. It is aimed at teams that want to replace manual spreadsheet workflows without building custom tools from scratch.
Wallie
Wallie is an open-source AI streamer that watches your screen, hears chat, and generates live commentary in a configurable persona. It runs locally on your machine with your own keys and is aimed at faceless content, autonomous streams, and real-time reactions.
Make Real
Make Real is a Next.js app that turns hand-drawn wireframes into working HTML prototypes using the tldraw SDK and AI models. It helps users sketch on a canvas, generate code, and review the result in an embedded preview.