Migrate from Flowise to Heym
Move Flowise chatflows and agentflows to Heym. Paste an exported flow, watch AI Convert rebuild it on a live canvas, then run it on your own infrastructure.
Flowise describes itself as an open-source, drag-and-drop builder for LLM chatflows and agent flows assembled from chains, tools, and vector store components.
When it is worth moving a workflow
The usual trigger is a chatflow that has to become an operation. A flow that answers questions well still needs a schedule, an approval step, a record of what each run cost, and a way to prove last Tuesday’s answer came from the documents somebody thinks it did. Those are runtime concerns, and they are easier to hold when the flow lives on the same canvas as the rest of the automation.
If you are still deciding rather than moving, the Heym and Flowise comparison covers the evaluation side with links to first-party documentation.
How the move works
- 1
Export from Flowise
Open the chatflow or agentflow in Flowise, export it as JSON from the flow menu, and copy the file contents.
- 2
Paste it into AI Convert
A safety check runs, then each construct is mapped to the closest native Heym node and streamed onto a live canvas while you watch.
- 3
Review, download, and import
Read the rebuilt workflow, download or copy it, import it into your own instance, and attach your credentials.
What each Flowise concept becomes
A reviewer's guide to the rebuilt workflow. Constructs without an exact equivalent are mapped to the closest Heym primitive, which is why the result is worth reading before it handles real traffic.
| In Flowise | In Heym | What changes |
|---|---|---|
| Chat Model component | LLM node | Model, temperature, and prompt carry over, with guardrails and structured JSON output on the same node. |
| Tool Agent | AI Agent node | The tool-calling loop carries over, extended with inline Python tools, MCP servers, and skills dropped onto the node. |
| Sequential Agents | Orchestrator agent with sub-agents | One agent takes a call_sub_agent tool and delegates to named agents on the same canvas, up to five levels deep. |
| Document Loader, Text Splitter, Embeddings, Vector Store | RAG / Vector Store node and the Vectorstores tab | Four components collapse into ingestion configured on one node, over Qdrant or Postgres with pgvector. |
| Retriever | RAG / Vector Store node in search mode | Semantic search with metadata filters and reranking, feeding results straight into an LLM or Agent node. |
| Memory components | Portal conversation history and agent context | Multi-turn history is held by the conversation surface, with automatic context compression as it grows. |
| Chains and connectors | Canvas edges | The wiring itself is the chain, and independent branches run in parallel without configuration. |
| Custom Tool (JavaScript) | Python tool on the Agent node | Custom tools are rewritten in Python. Review this one; it is the most likely place for a behavior difference. |
| Prediction API endpoint | Workflow API execute endpoint | Every workflow is callable over the API, and an agent workflow is also reachable as an MCP server. |
| Embedded chat widget | Portal | A workflow publishes as a chat UI with streaming, file upload, and optional per-user authentication. |
Check these before it goes live
- Re-ingest the documents. Vectors are rebuilt rather than copied, so load the same sources and spot-check retrieval before switching traffic.
- Rewrite and read any custom tool. Tools are Python here, and the conversion output deserves a line-by-line pass.
- Attach model and tool credentials in the Credentials tab.
- Re-test multi-turn behavior through the Portal, since memory is held by the conversation surface rather than by a component in the flow.
- Run one test execution and compare the answer against the same input on the source side.
Flowise migration FAQ
Can AI Convert read a Flowise export?
Yes. Paste the exported chatflow or agentflow JSON into AI Convert and it maps each component to the closest native Heym node, streaming the rebuilt workflow onto a live canvas you can download and import.
What happens to a custom tool written in JavaScript?
Agent tools in Heym are Python, defined inline on the Agent node, so a JavaScript custom tool is rewritten rather than transferred. Read the generated Python before the workflow handles real traffic.
Do my credentials come across?
No, and that is deliberate. Exported workflow files hold no API keys or secrets, so AI Convert rebuilds the structure only and you attach your own credentials in the Heym Credentials tab afterwards. Nothing copies a secret between two systems on your behalf.
Is the converted workflow ready to run?
Treat it as a reviewed starting point rather than a finished port. Open it on the canvas, attach credentials, run one test execution, and adjust any node that was mapped to the closest Heym primitive rather than an exact equivalent.
Can I run the result on my own infrastructure?
Yes. Heym is source-available under an MIT license with a Commons Clause and runs on your own servers with Docker or Kubernetes, so migrated workflows, their credentials, and their model calls stay inside your environment.
Moving from something else
AI Convert reads exports from 12 tools, and it also accepts a plain description of the automation when there is no file to export.