Why are Copilot’s responses in Teams better in the middle of the meeting than at the end?
When Copilot for Teams suggests a meeting summary usually 10 minutes before the end of the meeting with the standard prompt “Generate meeting notes,” the response is structured with bolded subtitles, very detailed, and linked to the transcription. Repeating the same prompt 10 minutes later returns a much lower-quality summary, without subtitles, very short. Furthermore, the summary specifies at the header “based on the transcript,” which the first prompt did not indicate.
Why does this happen? How can this be corrected as it makes Copilot inefficient?
When Copilot for Teams suggests a meeting summary usually 10 minutes before the end of the meeting with the standard prompt “Generate meeting notes,” the response is structured with bolded subtitles, very detailed, and linked to the transcription. Repeating the same prompt 10 minutes later returns a much lower-quality summary, without subtitles, very short. Furthermore, the summary specifies at the header “based on the transcript,” which the first prompt did not indicate.Why does this happen? How can this be corrected as it makes Copilot inefficient? Read More
