9.22System API as separate deployable app vs reusable subflow

This is an architectural question.

Suppose only one Mule application talks to Salesforce.

You could build:

customer-app
    ↓
salesforce-subflow

instead of deploying:

customer-process-api
      ↓ HTTP
salesforce-system-api

Why?

Because separate APIs introduce operational overhead.

A System API becomes more compelling when:

multiple consumers need it
independent lifecycle matters
central security/policy is useful
reusability matters
backend complexity deserves isolation