Workflow portability

Move AI workflows
without rebuilding them

Export a reviewed configuration bundle, then safely bring workflows, structured outputs, and pipeline dependencies into another ModelRiver project.

Recreating schemas and pipeline links by hand is slow and error-prone. ModelRiver moves the related configuration together and shows exactly what will change before import.

  • Keep workflows, schemas, and pipeline dependencies together
  • Preview conflicts and destination requirements before import
  • Create everything atomically, without partial migrations
  • Move configuration without exporting provider credentials or logs
Start free Read the documentation

From one project to another

A complete configuration transfer,
with a review step

Move the configuration that makes a workflow work together—without manually reconnecting its schemas and pipeline targets.

Source project

Export
W
support_triageWorkflow configuration
{ }
ticket_responseStructured output
→
classify → enrichPipeline dependencies

Versioned bundle

JSON · v1

workflow-bundle.json

2 workflows · 1 structure

Destination preview

Validate
✓
Names availableNo resources overwritten
✓
Dependencies linkedTargets and schemas resolved
✓
Project readyModels, providers, and quotas checked
Configuration collected
Bundle created
Preview passed
Ready for atomic import

Complete context

Dependencies included

Workflow exports include attached schemas and transitively required target workflows.

Review first

Preview before writes

See conflicts, unavailable providers, invalid schemas, and unresolved references early.

All or nothing

Atomic import

If validation or conflict checks fail, no workflows or structured outputs are created.

What moves

Export the right bundle for the job

Choose the resource you want to move. ModelRiver includes the configuration required to make that export useful in another project.

Workflow export

Move an interconnected workflow

Includes the selected workflow, its attached structured output, and every target workflow required by backend-pipeline events.

Best for: moving a complete feature without rebuilding its links.

Structured-output export

Share a schema on its own

Includes the selected structured output and its example without pulling in workflows that currently use it.

Best for: reusing response contracts across projects or teams.

Authoritative preview

Resolve issues before confirmation

The destination project is checked for name conflicts, quotas, provider connections, model availability, schemas, and references.

Think of it as: a preflight check for configuration.

Versioned JSON

Readable and reviewable

The version 1 bundle is plain JSON, so teams can inspect what it contains and keep independent dependency groups portable.

Important: exports can contain sensitive instructions and examples.

How it works

A clear path from bundle to destination

Use the same guided import flow from either the Workflows or Structured Outputs page.

  1. 01

    Open Import

    Start in the destination project from Workflows or Structured Outputs.

  2. 02

    Add your JSON

    Upload a version 1 file or paste its contents into the editor.

  3. 03

    Review the preview

    Inspect resources, dependencies, warnings, and field-level errors.

  4. 04

    Confirm safely

    Create the whole reviewed bundle together—or create nothing.

Built for controlled sharing

Portable configuration, not portable secrets

Treat an export as reviewed project configuration. Credentials and runtime data are never included; configure destination providers and permissions separately.

Included when relevant

  • Workflow settings and fallback configuration
  • Structured-output schemas and examples
  • Backend-pipeline events and target relationships
  • Descriptions, system instructions, and customer-field names

Always excluded

  • Provider credentials and API keys
  • Budgets and spend limits
  • Logs and runtime request data
  • Destination-project permissions and connections

Import permissions and safety. Organization owners and admins can import. Existing workflow and structured-output names are never overwritten, merged, or silently reused.

FAQ

Common questions about workflow portability

What to expect when moving workflows and structured outputs between projects.

Does importing overwrite an existing workflow?

No. A conflicting normalized workflow or structured-output name stops the import. ModelRiver does not overwrite, merge, rename, or silently reuse existing resources.

Are API keys or provider credentials exported?

No. Exports contain workflow configuration, not provider credentials, project permissions, logs, budgets, or runtime request data.

What happens if one resource in the bundle is invalid?

The import is rejected before any records are created. Imports are atomic, so a failed validation or conflict check cannot leave a partial migration behind.

Can I inspect the format before exporting my own workflow?

Yes. The documentation includes a complete downloadable version 1 example with workflows, structured outputs, fallbacks, customer fields, and a backend pipeline.

Move workflow configuration with confidence

Build once, review the bundle, and bring the configuration your next project needs—without recreating every schema and pipeline connection by hand.

Start freeView the import & export guide