AI-102 Question 426
Select 3Your company is building a customer support chatbot using Azure OpenAI Service. The chatbot must generate human-like responses based on customer queries and adhere to your company's brand tone. Additionally, the chatbot should integrate with your existing CRM system to fetch customer-specific data in real-time. Which of the following steps should you take to implement this solution?
- A
Use the Azure OpenAI Service to fine-tune a pre-trained GPT model with company-specific data.
- B
Implement the Azure Cognitive Search service to index and provide relevant customer support documents.
- C
Integrate the CRM system using Azure Logic Apps to fetch customer-specific data dynamically.
- D
Train a model from scratch using Azure Machine Learning to ensure it strictly follows the company's brand tone.
- E
Deploy the Azure OpenAI model to an Azure Kubernetes Service (AKS) cluster to handle high traffic.
Show answer and explanation
Correct answers: A, B, C
Explanation
To implement a generative AI solution for a customer support chatbot, fine-tuning a pre-trained GPT model ensures the chatbot is aligned with your company’s tone and context. Azure Cognitive Search allows the model to access relevant support documents, while integrating the CRM system with Azure Logic Apps enables customer-specific data retrieval. These steps ensure an efficient and personalized chatbot solution.
- A. Correct.
Fine-tuning a pre-trained GPT model with company-specific data ensures the chatbot generates responses that align with your company's tone and context.
- B. Correct.
Using Azure Cognitive Search helps the chatbot retrieve and use relevant support documents for more accurate and helpful responses.
- C. Correct.
Integrating the CRM system with Azure Logic Apps allows the chatbot to fetch real-time, customer-specific data, enhancing personalization.
- D. Incorrect.
Training a model from scratch is unnecessary and resource-intensive when pre-trained models like GPT can be fine-tuned effectively.
- E. Incorrect.
Deploying to AKS is not a mandatory step for handling high traffic; Azure OpenAI Service is scalable and can manage traffic without requiring AKS.