Prompt AI, ChatGPT and Azure AI to automatically update classification data based on customer feedback
Use Case:
Retail, Customer Service, Operations business that has need to classify each customer transaction and convert into actionable sentiment.
Business Value:
- With huge volume of data, utilize AI/GPT models to automatically process the collected customer sentiment and transform it into classified state so follow up actions are advised.
- Utilize built-in data classification prompt for AI model to transform the data and update the records.
Field Use Cases:
- Retail Industry: To track and classify customer feedback and work on next best actions.
- Others Customer services and support requires the organization to track customer sentiment and classify the customer feedback for upselling, case management and market latest offers.
Solution Stack:
Microsoft Power Platform | AI Hub |GPT |Power Automate | Prompt Customization
Sample File:
Steps:
- Got to make.powerapps.com»AI Hub»Choose Create text with GPT using a prompt AI model
- Ensure your Power Platform environment is in the region that supports these features. In this case, will be using USA/USA_Preview Environment
- Create custom prompt and name it as you like, in this case it’s called as AI Sentiment Analysis
- You will need an input variable, which could be customer sentiment or emails or updates.
- Save the prompt and click on the Test custom prompt button.
- You will need to provide input samples to evaluate your scenario.</li
- Detailed and specific inputs will yield better results through AI prompt.
Only a sample text input applied here. We can utilize automation functions for document input as well.
Click on test prompt and AI prompt model will start the action and provide you with the results.
Next Steps:
Auto update data post prompt process the classification. How will it apply for tons of record and to automatically update the data is where the combination of the solution stack using Power Automate helps with.
- The results are available for you to review, approve and take the next best actions.
- You can share the prompt with other users for reference. However, prompt in an app will be more meaningful and productive for end users.
Using AI Prompt as GPT generated text output in Power Apps for users to consume.
Create an automated flow to update records. In this example we will be using retail customer sentiment datasheet in //onedrive. The table has the below data with AI Output empty which is what the flow will update by calling AI Builder Prompt and ChatGPT text classification.
Results:
Below is the output post running the power automation flow. It took about 5-10 secs to update 1500 records. This can be particularly useful in the context of huge data volume and automated classification of data for other similar use cases.