Unanswered

Copilot Studio slow to respond to prompts

(1) Share

Report

Posted on by BizAppBee

Hi everyone, We have created a bot in Copilot Studio pointed to a Sharepoint site and it is properly answering questions however it is taking a really long time to respond to prompts. Any tips on how to improve the performance?

Any help is greatly appreciated. Thanks!

Categories:

General topics

I have the same question (0)

All responses (4)

Answers (0)

Audrie101 on at

Like (1)

Report

Thank you for raising this question. May I ask if all of your queries are directly coming from the end user of the chatbot? Or are they also coming from Power Automate flows or custom AI Builder GPT Models?

Have you added the SharePoint sites to the new "Knowledge" tab to see if this improves performance as well?
Microsoft Build 2024: Create custom copilots from SharePoint - Microsoft Community Hub

Thank you for your patience, and continued feedback.

Was this reply helpful? Yes No
CU20080144-0 on at

Like (1)

Report

Agreed, it is ridiculously slow. No, waiting 30-60 seconds for a one sentence answer is not helpful when I can create an assistant with OpenAI that works 10x faster than this.

Have you by chance found a solution to this? I'm definitely telling my company to move on and try something better.

Was this reply helpful? Yes No
Mahesh Chintha 158 on at

Like (1)

Report

We also see ~30 seconds latency when the knowledge source is SharePoint irrespective of classic or modern Knowledge source. Use the latency message for end users that response of being generated.

We observed better performance of ~20 seconds with the US based Power Platform tenant/environment. Check the network trace if there are any proxys/multiple hops involved. MS says ~20 secs is acceptable with GPT4 and SharePoint. If this is not acceptable you can upload documents directly to Copilot Studio for faster responses.

Was this reply helpful? Yes No
Suggested answer

CU28031904-0 2 on at

Like (0)

Report

Hi,

Microsoft Copilot Studio is a PaaS (Platform as a Service) application, it automatically scales based on usage. As you consistently interact with it and usage increases, the platform dynamically allocates more resources, ensuring better performance over time. This means that higher engagement leads to improved responsiveness and efficiency, as the system optimizes itself to handle growing demand effectively.

So basically to get better response times, try to test writing messages consistently. In that case you will get better performance.

Was this reply helpful? Yes No