Generative AI Leader Practice Question: Integrate GenAI into their existing customer…

Answer analysis

Option-by-option breakdown

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

• ✓

  API-first integration by calling Vertex AI API from the CRM's custom code

  Why this is correct

  A third-party SaaS CRM cannot host custom model runtimes, so API-first integration calling the Vertex AI API from the CRM's custom code is the only viable pattern. It satisfies the constraint of extending an externally managed application without access to its underlying infrastructure.

• ✗

  Building a standalone AI application and exporting data manually

  Why it's wrong here

  Manual export breaks the real-time, in-context integration the CRM requires, adding latency and human error. It tempts as a quick proof of concept with no vendor API work, yet the correct pattern calls the CRM's APIs or embeds GenAI within it, keeping data synchronised automatically rather than batch-copied.

• ✗

  Embedding GenAI using Google Workspace add-ons

  Why it's wrong here

  Workspace add-ons extend Gmail, Docs and Calendar, not a third-party CRM's own UI or data model. They tempt because they are Google-native and low-code, but the correct pattern uses the CRM's supported extension or API surface, which add-ons cannot reach inside a non-Google SaaS application.

• ✗

  Using Apps Script to extend Google Sheets connected to the CRM

  Why it's wrong here

  Apps Script extending Sheets only reaches data the CRM exports into that sheet, not the CRM's live records or interface. It tempts because Sheets and Apps Script are familiar and free, but the correct pattern integrates through the CRM's own API or extension framework, preserving real-time access.

Go deeper

Related to this question