Why are the agents performing so poorly when analyzing data?

(0) Share

Report

Posted on by JG-03120804-0

Hello,

i have started experimenting with agents in Copilot Studio. Using small datasets, 1,000 records, csv, on OndeDrive or on Sharepoint. Despite providing instructions and sample questions, the agent is not able to answer even the basic questions. It looks like the agent is taking into account only a fraction of the data on OneDrive for example. I have also tried Azure Databricks connections, with the same poor results. Does anyone have similar experiences, or can anyone provide insight into the proper settings i should use? any input is welcome. Regards, Jeffrey

Categories:

General

I have the same question (0)

All responses (1)

Answers (0)

Yacine 7 on at

Like (0)

Report

Why are the agents performing so poorly when analyzing data?

Hello Jeffrey,

I’m facing the same issue. I don’t know why, and I tested a use case with only a sample of 10 rows. :(
The chatbot is hallucinating, sometimes it responds correctly, and sometimes it doesn’t. But overall, the success rate is definitely not high.

I prefer to use Azure AI Services, it's obviously better for me.

Best regards,

Yacine

1 people found this reply helpful.

Was this reply helpful? Yes No