How can I improve the capabilities of my Copilot bot to match those of Azure OpenAI Studio's Chat playground?

Federico 5 Reputation points
2024-02-28T08:32:03.78+00:00

We deployed a couple of Power Virtual Agents (now Copilots) by following the Azure OpenAI Studio -> Copilot Studio flow. Because of this, we tested the bots with our data (Azure AI Search) in both Azure OpenAI Studio and Copilot Studio. There are certain differences between the tests in both products that after some reading, I was able to identify as current limits for Copilots. For example Copilots are not turn-based bots. However testing the bot in Azure OpenAI Studio's Chat playground proves that in there turn-based bots are possible. I noticed as well that if we ask to retrieve data and present it in a table format, the Chat playground has no problems doing so but Copilots will respond with "I'm sorry, I'm not sure how to help with that. Can you try rephrasing?". Is there some work being done for Copilots answers to eventually reach the same level as Azure OpenAI? Is there something I can do to improve the situation? I considered the option of using Azure Bot Services instead which might yield the results I'm looking for, but I would be trading the option of carrying out automated tasks. Something that Copilots can do.

Microsoft Copilot
Microsoft Copilot
Microsoft terminology for a universal copilot interface.
282 questions
0 comments

1 answer

Sort by: Most helpful
Most helpful Newest Oldest
  1. Shagupta Maniyar-MFST 0 Reputation points Microsoft Vendor
    2024-09-18T11:30:37.1233333+00:00

    It sounds like you've encountered some of the current limitations of Copilots compared to Azure OpenAI Studio's Chat playground. While Copilots are designed to assist with a wide range of tasks, there are indeed some differences in capabilities, especially when it comes to turn-based interactions and presenting data in specific formats like tables.

    Here are a few suggestions to help improve the situation:

    1. Enhance Prompts: Sometimes, rephrasing the prompts can help Copilots understand and respond better. For example, instead of asking directly for a table, you can guide the Copilot by specifying the format you expect in more detail.
    2. Use Azure Bot Services: As you mentioned, Azure Bot Services might provide the flexibility you need for turn-based interactions and more complex data presentations. While you might trade off some automated task capabilities, you can still integrate Azure Bot Services with other tools to achieve a comprehensive solution.
    3. Combine Services: Consider using a combination of Copilots and Azure Bot Services. You can leverage the strengths of each platform by using Copilots for automated tasks and Azure Bot Services for more complex interactions. This might require some integration work but can provide a balanced solution.
    4. Stay Updated: Microsoft continuously works on improving Copilot capabilities. Keep an eye on updates and new features that might address some of the current limitations. You can follow the official Microsoft blogs and documentation for the latest information123.
    5. Provide Feedback: If you encounter specific issues or limitations, providing feedback to Microsoft can help prioritize these improvements. Your insights can contribute to future updates and enhancements.

Your answer

Answers can be marked as Accepted Answers by the question author, which helps users to know the answer solved the author's problem.