Yeah, this is one of those confusing areas where Copilot/agent flows and Power Automate look similar but behave quite differently.
What’s happening is the action you’re using:
is asynchronous. It only starts the conversation and returns a ConversationId, but it doesn’t wait for or return the agent’s response. That’s why you’re not getting any output to post or email.
Also, at the moment, there isn’t a supported way to take that ConversationId and fetch the response later via Power Automate. So that path won’t help here.
If you want the actual response (so you can post it in Teams or send an email), you need to switch to a pattern where the flow waits for the agent to respond.
The right approach is to use:
Now, you mentioned you need different prompts for BoD and EoD. That’s totally doable.
Instead of hardcoding the prompt, just pass it dynamically from your recurrence flow:
Example:
and for the second flow:
These actions will return the actual response from the agent, which you can then use in:
- Microsoft Teams → Post message in chat
- Outlook → Send email
For example:
If you prefer to keep your current “Send prompt” action, then the only working pattern is to let the agent handle the response itself:
- Inside the agent, create a topic that:
- Handles the prompt (BoD/EoD)
- Calls a Power Automate flo
- That flow then:
- Sends email
- Or posts to Team
But in this case, the trigger must originate from the agent conversation, not just an external recurrence flow. That’s why your “secondary flow” likely isn’t firing.
So in short:
- “Send prompt to Copilot” → fire-and-forget (no response available)
- “Execute Agent / Execute and Wait” → returns response (best for your use case)
- ConversationId → cannot be used to retrieve response
For your scenario (scheduled BoD/EoD summaries), the cleanest setup is:
- Recurrence trigger
- Execute Agent (with dynamic prompt)
- Use returned response → send to Teams or email
Best regards,
Satyam Pandey
✅ If this helped solve your issue, please mark it as Accepted Solution so others can find it quickly.
❤️ If it was helpful, please click Yes on “Was this reply helpful?” or give it a Like.
🏷️ For follow-ups, feel free to tag @TechFreak
💼 LinkedIn
▶️ YouTube