Power Platform Community Forum Thread Details

Hi Guys ,

I am in desperate need of help, I have experience with Power Automate flows, and model driven apps , but my knowledge in copilot studio is still little. I have a model driven app that i am making for procurement process at my organization , Part of the process is that users upload three qoutations in the notes timeline. I wanted to utilize the builtin interactive agent in model driven apps , whereby the agent will analyze those three quotations and give a recommendation on the best one it would suggest and display this using Agent Response (Preview Component) . I need help with analyzing the current steps i had done.Is it the right way , or there are steps I am not doing well.

Categories:

Experimental features

@CU18062040-0

Integrating Copilot Studio with your model-driven procurement app to analyze uploaded quotations and provide intelligent recommendations is a powerful use case. Let’s walk through what you’ve done and what you might need to adjust or add.

✅ What You’ve Done So Far (Great Start!)

Built a model-driven app for procurement.
Users upload three quotations in the timeline (notes).
You want to use the built-in interactive agent to:
- Analyze these quotations.
- Recommend the best one.
- Display the result using the Agent Response (Preview Component).

🔍 Key Considerations & Next Steps

1. Accessing Uploaded Quotations

The timeline (notes) stores attachments in the Annotation table. To analyze them:

Use Power Automate or a plugin to retrieve the attachments from the related record.
Ensure the agent has access to the content of the files (PDF, DOCX, etc.).

2. Copilot Studio Agent Setup

To analyze and respond intelligently:

Create a Copilot Studio agent with a custom topic like “Analyze Quotations”.
Use Power Fx or Azure OpenAI (if enabled) to process the file content.
You may need to pre-process the files using Power Automate and pass the extracted text to the agent.

3. Agent Response Component

Embed the Agent Response (Preview) component in your form.
Trigger the agent when the user clicks “Analyze Quotations”.
Display the recommendation (e.g., best price, delivery time, vendor rating).

4. Data Flow Example

User uploads 3 quotations → Power Automate extracts text → Sends to Copilot Studio agent → Agent analyzes and recommends → Response shown in Agent Response component

🛠️ Suggestions to Improve Your Setup

✅ Use Power Automate to extract and clean text from attachments.
✅ Pass structured data (e.g., price, vendor name, delivery time) to the agent.
✅ Train your agent with sample quotation formats so it knows what to look for.
✅ Log the recommendation in a custom table for audit/history.