> For the complete documentation index, see [llms.txt](https://doc.wearepatchworks.com/product-documentation/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://doc.wearepatchworks.com/product-documentation/ai/patchworks-ai-studio/platform-agents/migration-agent.md).
# Migration Agent
The Migration Agent rebuilds integrations from other platforms in Patchworks.
{% hint style="success" %}
The Migration Agent is **active** on your account. Access it from the AI Studio.
{% endhint %}
Upload an export of a flow from your current platform - Celigo, Boomi, Workato, Make, Zapier, or any other tool that can export its configuration - and the agent works out what the integration does, then hands a build brief to the Implementation Agent, which creates the process flow(s) for you.
Rather than documenting each step of an old integration by hand and rebuilding it shape by shape, you get a plain-language breakdown of what the flow does, a clear list of what the export doesn't tell us, and a working starting point in Patchworks.
#### What it does
The Migration Agent does the analysis; the [Implementation Agent](/product-documentation/ai/patchworks-ai-studio/platform-agents/implementation-agent.md) does the building. It takes a required file upload, reads it, and passes on a brief. When you upload an export, the agent works through it and returns:
* **Platform and flow** - which system the export came from, plus the flow's name, schedule and timezone.
* **Source and destination** - where data is read from and written to, and how (API, RESTlet, file transfer, database).
* **Steps, mappings and logic** - every step in order, with the field mappings, conditions, filters and hardcoded values it can find.
* **What it can't know** - credentials, saved search IDs, lookup tables and mappings that aren't in the export, listed so you can complete them after the build.
* **A test plan** - the checks that confirm the new flow matches the old one, including error scenarios.
Because every platform exports differently, the agent works from what the file contains rather than a fixed template. Where something is unclear, it says so rather than guessing, and it asks you before anything is built.
#### How the export is translated
| In the export | In Patchworks |
| ---------------------------------- | ---------------------------------------------------------------------- |
| Trigger or schedule | Trigger shape (schedule, webhook or event listener) |
| Source and destination shapes | Connector shapes using your installed connectors |
| Field mappings and transformations | Map shape, with mappings carried across where the export includes them |
| Conditions and branching | Filter, Route and Branch shapes |
| Lookups and reference tables | Cross-reference lookups |
| Error handling and alerts | Try/Catch and Notify shapes |
| Custom code | Script shape |
#### How to use it
The Migration Agent is available from the **AI Studio** page under **Platform Agents**, and from the **AI Assistant**.
1. **Open the Migration Agent** - Navigate to **AI Studio → Platform Agents**, select the Migration Agent card and click **Start conversation**.
2. **Upload the export** - Attach the export file from your current platform. This is required, and the agent will not start without one. Attach the original export (usually JSON, XML or a zip) rather than a screenshot, and add anything the file won't show, such as a sample payload or a note on what the flow is for. Then tell the agent what you want - for example, "Convert this flow into a Patchworks process flow."
3. **Review the analysis** - The agent identifies the platform, flow, schedule, source, destination and steps, and lists any open questions. Answer them in the conversation - the more it knows, the more complete the build.
4. **Confirm the build** - The agent tells you exactly what it can create and what you will need to configure yourself, then asks whether to proceed. Confirm, and it hands the brief to the **Implementation Agent**, which creates the process flow - or several flows, if the export contains more than one - and reports back with what was built and a link to each flow.
5. **Finish the configuration** - Open the flow and complete the parts the export couldn't provide: connector instances and credentials, saved search or query IDs, cross-reference lookups, and any mappings flagged for the mapping editor. Run it against test data before enabling it.
> A migrated flow is a faithful starting point, not a finished integration. Connections and credentials never travel in an export, and mappings are only as complete as the file that contained them - so always review the flagged items and test before switching over from the old platform.
