Question

A SaaS company receives customer support tickets in over 15 languages. The product team wants a low-code solution to automatically extract mentioned product names, organizations, and locations from incoming ticket text so tickets can be routed to the right regional team. They have no labeled training data and want to ship quickly using a managed Google Cloud service. Which approach best fits these requirements?

Accepted Answer

Use the Cloud Natural Language API entity analysis feature to extract entities directly from the ticket text without training a custom model.

Answer

Train an AutoML text entity extraction model in Vertex AI using the historical tickets after manually labeling a representative sample.

Answer

Fine-tune a Gemini model in Model Garden on the multilingual ticket corpus to perform named entity recognition.

Answer

Use Document AI's Form Parser to extract entity fields from the ticket text across all supported languages.