Migrate flows from n8n, Make or Zapier
If you already have automations in n8n, Make or Zapier, you don’t need to rebuild them by hand. Export the flow, paste it into the assistant chat and it suggests how it will look in NappAI, builds it with you and tests it.
1. Export the flow
| Tool | What to export |
|---|---|
| n8n | The workflow as JSON (workflow menu → Download, or select the nodes and copy them). |
| Make | The scenario’s blueprint (the scenario’s ⋯ menu → Export Blueprint). |
| Zapier | The JSON export of your Zaps, from your Zapier account settings. If it contains several Zaps, each one becomes a separate flow. |
2. Paste it into the chat
Paste the whole JSON into the message box, as it is, and add what you expect if you like (“move it to NappAI and send the alert through Telegram instead of Slack”). It can have text around it: the assistant recognizes it.
A message with an export can be much longer than a normal message, so a large flow fits.
3. Review the proposal
Before the assistant replies, a summary of what it has read appears: the tool, the flow’s name, the number of steps and how it can be built in NappAI. Below it, a table with one row per step of the original flow:
| Column | What it shows |
|---|---|
| Step | The name of the original step. It marks which one is the trigger and which ones were disabled. |
| What it is | What that step did. |
| In NappAI | The component that replaces it (or several candidates). “PyCode” if there is no component for that service and custom code will be needed; that there is no equivalent and it has to be decided with you; or that the step isn’t needed in NappAI. |
Next, the assistant explains what it chooses where there are several candidates, what changes compared with the original and what it needs to ask you. It doesn’t create anything until you are happy with it, and always as a new flow in the folder you choose.
What changes when you migrate
- Credentials aren’t imported. The keys and passwords from the original flow are removed from the export. For each service you choose a credential you already have in NappAI or create one, just like in a new flow.
- The webhook URL changes. If the flow started with a webhook, you will have to point the source service to the new NappAI address.
- Scheduling is separate. In NappAI how often a flow runs isn’t a component: the flow is scheduled in the Scheduler when it is saved. Make doesn’t export a scenario’s schedule, so the assistant will ask you for it.
- Services without a component are handled with code. If NappAI has no component for a service (for example, HubSpot or Pipedrive), the assistant solves it with a Python Code component that calls its API, and tells you so.
- Some logic behaves differently. A Make router runs every route whose condition is met; in NappAI a branch is “if / else”, so the assistant will confirm what you want. And a condition whose branches join again doesn’t come back together if one of the branches produces nothing: if that is your case, the assistant solves it another way.
4. Test it and turn off the original
The assistant tests the new flow with sample data, branch by branch too, and saves those tests as the flow’s test cases. When it finishes, it sums up what it has built and what is left in your hands.