Generative AI Leader Practice Question: A developer needs to integrate a GenAI model into…

Answer analysis

Option-by-option breakdown

For each option: why learners choose it and why it is or isn't the right answer here.

• ✗

  Use Apps Script to call the model from Google Sheets

  Why it's wrong here

  Apps Script runs only within Google Workspace, so it cannot reach an on-premises CRM's REST APIs or serve as its integration layer. It suits automating Sheets, Docs and Gmail tasks. Here the CRM itself must call the model through its own REST endpoints.

• ✗

  Fine-tune a model on CRM data and deploy it on-premises

  Why it's wrong here

  Fine-tuning alters model weights for behaviour or domain adaptation; it does not connect the CRM to the model, and on-premises deployment of a large model is unnecessary here. Fine-tuning suits specialised accuracy needs. The requirement is a REST integration between CRM and model.

• ✓

  Implement an API-first integration by calling Vertex AI API from the CRM's backend

  Why this is correct

  Calling the Vertex AI API from the CRM's backend preserves the on-premises REST architecture, so no inbound connectivity or data migration is required. The API-first pattern lets the existing CRM orchestrate model inference server-side, satisfying the on-premises constraint while keeping credentials and customer data within the trusted boundary.

• ✗

  Build a custom Google Workspace add-on

  Why it's wrong here

  A Workspace add-on surfaces UI inside Gmail, Docs and Calendar; it cannot integrate an on-premises CRM's REST APIs. Add-ons suit extending Workspace user interfaces. This scenario needs the CRM's own REST endpoints calling the model directly.

Go deeper

Related to this question