MC1239826 High
Microsoft Copilot Studio – Create Inputs from Existing Conversations
Summary AI-generated
Copilot Studio will let users create evaluation inputs from real conversations, supporting improved test sets and agent performance; available March 31, 2026.
Written by Azure OpenAI (gpt-4.1) from the text of the post below. It can be incomplete or wrong; the original post is authoritative.
Similar posts
Search for more like this- MC1165633 Microsoft Copilot Studio – Draft with Copilot
- MC1325011 Microsoft Copilot Studio - Evaluate the entirety of multi-turn conversations
- MC1262498 Microsoft Copilot Studio - Analyze user sentiment from agent conversations
- MC1403399 Microsoft Copilot Studio - Create agents optimized for Microsoft 365 and Microsoft 365 Copilot users
- MC1259433 Power Apps - Use Microsoft 365 Copilot in model-driven apps
- MC1323229 Microsoft Copilot Studio - Automate web and desktop apps with computer use
Original post from Microsoft
We are announcing the ability to create inputs from existing conversations in Microsoft Copilot Studio. This feature will reach general availability on March 31, 2026.
How does this affect me?
This feature enables users to generate evaluation inputs directly from real production conversations by converting past customer interactions into reusable structure inputs. By using actual customer interactions, users can create evaluations that reflect real usage patterns, uncover quality gaps, validate fixes, and improve agent behavior.
This feature supports generating test sets from themes, question lists, or reactions, and automatically selects up to 100 recent questions when working with large datasets.
What action do I need to take?
This message is for awareness, and no action is required.
How does this affect me?
This feature enables users to generate evaluation inputs directly from real production conversations by converting past customer interactions into reusable structure inputs. By using actual customer interactions, users can create evaluations that reflect real usage patterns, uncover quality gaps, validate fixes, and improve agent behavior.
This feature supports generating test sets from themes, question lists, or reactions, and automatically selects up to 100 recent questions when working with large datasets.
What action do I need to take?
This message is for awareness, and no action is required.