web
You’re offline. This is a read only version of the page.
close
Skip to main content
Community site session details

Community site session details

Session Id :
Copilot Studio - Autonomous Agents
Suggested answer

Inaccurate RAG response

(1) ShareShare
ReportReport
Posted on by 4
Hi all,
 

I’ve been testing Copilot but it isn’t working as intended. The main issue is that it keeps providing inaccurate responses, even though I’ve tried feeding it knowledge from SQL tables, Excel files, and CSV files. On top of that, it’s unable to retrieve information from a SharePoint list. I’m currently using the GPT-5 model, but I haven’t seen any improvement. I’ve also experimented with Power Automate flows to try and retrieve the data in a different way, but the results are still not reliable. At this point, it feels like I’ve tried everything, but Copilot Studio just isn’t delivering the level of accuracy or integration needed.

Is anyone else having the same experience, or have any reccomendations?
 
Thanks,
I have the same question (1)
  • Suggested answer
    Michael E. Gernaey Profile Picture
    49,350 Super User 2025 Season 2 on at
    Inaccurate RAG response
     
    Its unfortunately very hard to diagnose a copilot issue with just that description.
     
    I am not sure what you mean you cannot retrieve information from SharePoint (in what way, lists, files?)
     
    Please outline your Bot, the instructions you are using, the DataSources, the Settings you are using for Gen AI, Public websites, you name it. what channels are you posting it to (Teams/M365?), what is your authentication set too?
     
    If you look in the Activities tab in the designer what is it doing or not doing.
     
    Please give a really solid example of:
    -in my SQL Table, I have a field called Blue and it contains a simple true false, it also has an ID number for an order, and when I ask if my order is Blue or not, it says, Go jump in a lake, or has no answer.
     
    So we can figure out via the Activities tab and your configuration what it thinks its supposed to do, versus what its trying.
     
    Are you using Topics, Tools, in your instructions are you telling it what sources to use for specific questions?
     
    Happy to help but need more information as beyond that its not enough to understand as you also forgo to give an example of what you asked it and what it said or didn' say.
     

Under review

Thank you for your reply! To ensure a great experience for everyone, your content is awaiting approval by our Community Managers. Please check back later.

Helpful resources

Quick Links

Responsible AI policies

As AI tools become more common, we’re introducing a Responsible AI Use…

Tom Macfarlan – Community Spotlight

We are honored to recognize Tom Macfarlan as our Community Spotlight for October…

Leaderboard > Copilot Studio

#1
Michael E. Gernaey Profile Picture

Michael E. Gernaey 390 Super User 2025 Season 2

#2
Romain The Low-Code Bearded Bear Profile Picture

Romain The Low-Code... 273 Super User 2025 Season 2

#3
chiaraalina Profile Picture

chiaraalina 91

Last 30 days Overall leaderboard