ModelRiver Help Center
On this page

What's the difference between a project and a workflow?

Updated

A project is a container: it holds your provider keys, API keys, workflows, webhooks and request logs. A workflow is one named AI setup inside a project (provider, model, backups, instructions, output schema) that your app calls by name.

What goes into a workflow?#

  • The provider and model, plus up to two backup models
  • System instructions
  • An optional structured output (JSON schema)
  • Optional caching and conversation memory
  • An optional event name, if your backend should process the result before it reaches the user

Your code only knows the workflow's name, for example support_reply. Change the model or prompt in the console, publish it, and your app picks it up without a deploy.

How should I organize projects?#

Use separate projects for things that should stay apart:

  • Environments: one project for production, one for staging. Each has its own keys, so a staging key can never touch production.
  • Products or clients: separate projects keep logs, limits and keys separate.

Keep related features in one project as separate workflows.

Is there a limit on workflows?#

No. Create as many as you need. Each request counts toward your plan whichever workflow it uses.

Can I rename a workflow?#

Not after it's created, because your code calls it by name. Create a new workflow with the new name, switch your code over, then delete the old one.

Learn more#

Was this article helpful?