June 13, 2026Ceren Kaya Akgün
Migrate to Heym from n8n, Flowise, Dify & More
Move your n8n, Flowise, Dify, Langflow, sim.ai, or Activepieces workflow to Heym. Paste an export into AI Convert, preview on a canvas, and import.
You built real workflows in n8n, Flowise, Dify, Langflow, sim.ai, or Activepieces. Now you want to move them to Heym, and the thing standing in your way is the thought of rebuilding every node by hand. This guide shows the shortcut: paste your exported workflow into AI Convert, a free tool, preview the rebuilt workflow on a canvas, and import it into Heym. No manual rebuild.
The migration problem nobody solves
Search for how to move a workflow off n8n and you find the same advice everywhere: export the JSON, then rebuild it somewhere else from scratch. The export is easy. The rebuild is the wall.
This is not a niche complaint. The n8n community has open feature requests asking to convert workflows to code and to export a workflow to another language. People who already invested in building automations want a way to take that work with them, and the platforms they built on do not offer one. The export format is a backup, not a migration path.
Heym takes the other side of that request. AI Convert reads a workflow you already have and rebuilds it as a Heym workflow you can run, so the hours you spent designing logic in another tool are not thrown away when you switch.
What AI Convert does
AI Convert is a free tool at heym.run/convert. You give it one of two things: a workflow export from another platform, or a plain-English description of the automation you want. It runs a safety check, then generates a Heym workflow and streams it onto a canvas so you watch the nodes and connections build in real time. When it finishes, you download the workflow as JSON or copy it, then import it into Heym.
It maps each source concept onto the closest native Heym node. A trigger becomes an Input, Cron, or IMAP trigger. A model step becomes an LLM or Agent node. An HTTP request stays an HTTP node. Loops, conditions, and outputs carry over as their Heym equivalents. The result is a working starting point that you review and adjust, not a black box.
How your nodes map to Heym
You do not need to memorize the mapping, because AI Convert handles it. Still, it helps to know where each piece lands. This table shows how the common building blocks in n8n, Flowise, Dify, Langflow, sim.ai, and Activepieces translate into native Heym nodes.
| What your workflow does | In n8n, Flowise, Dify, Langflow, and more | Heym equivalent |
|---|---|---|
| Start on an event | Webhook, schedule, email, or chat trigger | Input (webhook), Cron, IMAP Trigger |
| Call a model | LLM, OpenAI, or model node | LLM node |
| Run an agent with tools | Agent or tool nodes | Agent node with tool calling, MCP, and skills |
| Retrieve from a knowledge base | Vector store plus retriever | RAG node over Qdrant |
| Call an external API | HTTP request node | HTTP node |
| Branch on a condition | IF, router, or switch | Condition, Switch |
| Loop over items | Loop, iterate, or split in batches | Loop node |
| Return a result | Respond or output node | Output node |
When a source node has no exact match, AI Convert picks the nearest Heym node and labels it so you can adjust it on the canvas. Triggers and credentials are the two areas to check first, since those are platform specific.
Migrating from n8n
n8n exports a workflow as a single JSON file. From the editor, open the workflow, use the menu to download or copy it, and you have the file AI Convert needs.
- Open heym.run/convert.
- Set the source tool to n8n.
- Paste the exported JSON into the box.
- Generate, then watch the workflow appear on the canvas.
- Download the result and import it into Heym.
Two things that trip people up when moving n8n workflows, and how they work here:
Credentials do not come across, and that is correct. An n8n export never contains your API keys, which is why credentials seem to disappear on import. AI Convert rebuilds the structure, and you attach your own credentials in Heym's Credentials tab afterward. Copying secrets between systems would be the unsafe option.
Version compatibility stops being your problem. A recurring n8n support theme is workflows breaking across versions on import. Because AI Convert rebuilds the workflow as native Heym nodes rather than replaying a version-specific file, you are not carrying that fragility with you.
If your goal is to recreate a workflow from a plain description instead of a file, that path is built in too, and it is the same engine described in text to workflow.
Migrating from Flowise, Dify, Langflow, sim.ai, and Activepieces
The flow is identical for the other tools. Export your project, set the matching source in AI Convert, paste, and generate.
- Flowise. Export the chatflow as JSON, set the source to Flowise, and paste it. Flowise chains of LLM and retrieval steps map cleanly onto Heym's LLM, Agent, and RAG nodes.
- Dify. Export the app or workflow, set the source to Dify, and paste it. Dify's app logic becomes a Heym workflow on the canvas.
- Langflow. Export the flow JSON, set the source to Langflow, and paste it. Langflow graphs translate naturally because both tools think in nodes and edges.
- sim.ai. Export your workflow, set the source to sim.ai, and paste it. The agent and tool steps become Heym Agent and LLM nodes.
- Activepieces. Export the flow, set the source to Activepieces, and paste it. Its triggers and action pieces map onto Heym triggers, HTTP, and integration nodes.
If a source has no export you can reach, skip the file and describe the workflow in plain English. AI Convert generates the matching Heym workflow either way.
After the conversion: a short checklist
A converted workflow is a draft you finish, not a finished product. Spend a few minutes on these before you rely on it.
- Read the canvas. Click through the nodes and confirm the logic matches what you expect. The preview shows every node, so nothing is hidden.
- Attach credentials. Add an API key in the Credentials tab for each node that calls an external service.
- Run it once. Trigger the workflow with a test input and check the output. Heym shows execution results inline so you can see where data flows.
- Adjust the near misses. If a node was mapped to the closest Heym primitive rather than an exact match, swap or reconfigure it on the canvas.
The import step itself is covered in the download and import flow, and the broader picture of building on a canvas is in what is AI workflow automation.
What migrating actually saves you
The reason to migrate is rarely the license fee. It is the bill that grows every month as your automations run more often. Three things drive that bill, and they are worth estimating before you move.
- How often each workflow runs. A flow triggered every minute costs far more over a month than one triggered once a day.
- How many steps each run executes. More nodes per run means more billed operations per trigger.
- How much data moves through it. Large payloads and high-volume retrievals add up.
Multiply your busiest workflow's runs per month by its step count and you have the execution volume you are paying cloud margins on. That number is usually where self-hosting starts to pay for itself.
Heym runs on your own infrastructure, so there is no per-execution margin stacked on top of your compute. You trade a usage-based bill for a predictable hosting cost, and the trade gets better the more your workflows run. Self-hosting is not free, so the win is largest for high-volume or high-step workflows, which are exactly the ones worth migrating first. Once you are on Heym, the AI agent cost optimization guide covers how to cut the model and runtime costs that remain.
Why teams move to Heym
Migration is only worth it if the destination is better for what you are doing. Three reasons come up most often.
Self-hosted and source-available. Heym runs on your own infrastructure. Your workflows and the data they process stay with you, and there is no per-execution cloud bill growing with your usage.
AI-native, not AI-added. Heym ships first-class LLM and Agent nodes with tool calling, multi-agent orchestration, built-in RAG, and MCP support. If your workflows are getting more agent-heavy, that is the core of the platform rather than a plugin on top of a trigger-action engine. For a wider view of the landscape, see the best AI agent builders in 2026.
Low friction to start. With AI Convert, the usual reason not to switch, the rebuild cost, mostly goes away. You bring your existing logic, attach your keys, and run.
Try it with one workflow
Migrate in phases rather than all at once. Pick a single workflow you already trust, ideally a high-volume one where the savings are clearest, export it, and run it through AI Convert. Five minutes later you have it on a Heym canvas, ready to import and run. Prove it on that one, then move the rest. If you do not have an export handy, describe a workflow you want and let it build one for you.
Start at heym.run/convert, and if Heym earns the switch, a GitHub star helps other people building workflows find it.
Founding Engineer
Ceren is a founding engineer at Heym, working on AI workflow orchestration and the visual canvas editor. She writes about AI automation, multi-agent systems, and the practitioner experience of building production LLM pipelines.
Enjoyed this post? Get the next one in your inbox.
A monthly note with practical ideas for building AI workflows that hold up in production. No noise, and you can unsubscribe anytime.