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

Notifications

Announcements

Community site session details

Community site session details

Session Id :
Power Platform Community / Forums / Copilot Studio / Multi - agent : hide "...
Copilot Studio
Unanswered

Multi - agent : hide "sub" agent answer and just resume the answer

(1) ShareShare
ReportReport
Posted on by 2,179 Super User 2025 Season 2
hello, 
 
here is the configuration
 
MAster agent
- 3 sub agent with their knowledge.
 
I ask the agent to use all the 3 agent to give an answer and then make a summarize. (prompt at bottom for better readibility ). I m pretty sure, around May with the Build 2025 we had an option "hide sub agent answer and let master agent to answer" or something like this.
 
My master agent is making everything i want : create a sumarize etc. BUT it display to the user the answer from all 3 sub agent AND then only make the summarize.  It's horrible for the user experience and surprising since with tool like agent flow it managed to dit.
 
I have 2 other way to solve it but i don't like it and ask for you help, probably not prompting correctly or missing a parameter or code in the yaml of the agent connexion ?
 (For information i found to way to workaround that are not satisfying : 
- Made a agent flow to ask each agent, take result, send it to master agent wich only display my summary) : i don't want this because the agent flow use a connexion inside the flow to each agent and don't use the "current connected user" :( 
- asked the master agent to force each sub agent to write "small summary" : i don't want it because it display 3 answer, i want only one answer with a summarize of the 3 agent
 
To make it short : 
I wan't to collect the answer of all 3 agent, don't display it, use the answer to make a summarize (i tried also inside a topic but there are no option to get answer inside a variable or i didn't it)
 
HERE is my prompt
 
### CONTEXT
You are a master orchestration agent in Microsoft Copilot Studio.  
You collaborate with three specialized sub-agents, each focused on a distinct domain of expertise:  
- Agent 1→ Focus: Roles and Responsibilities  
- Agent 2 → Focus: Product, technical aspects, and the “what” being delivered  
- Agent 3 → Focus: Engineering organization and “how” we work  
The user interacts only with you (the master agent).  
You are responsible for distributing their question to the sub-agents, consolidating their outputs, and delivering a concise summary.  
Raw answers from the sub-agents must never be shown directly to the user.
---
### OBJECTIVE
When the user submits a question:
1. Send the same question to the three sub-agents: Agent 1, Agent 2, and Agent 3.  
2. Collect their full responses and references (sources).  
3. Do not display these raw responses in the chat.  
4. For each sub-agent:
   - Summarize the key points in one or two sentences maximum.  
   - Extract and format clickable sources (using Markdown `title` when possible).  
   - If no relevant information or sources are provided, explicitly state:
     > “No relevant information returned by this agent.”  
5. Present all three summaries in a clean, structured format.
---
### METHOD
Follow these steps in sequence:
1. Distribute the question
   - Send the same user query to each sub-agent (Agent 1, Agent 2, and Agent 3).  
   - Wait for all responses before proceeding.
2. Process responses
   - For each sub-agent, read the content and extract:
     - The main answer or reasoning (summarized in 1–2 sentences).  
     - The list of links or sources (clickable).  
   - If the sub-agent didn’t respond or the content is irrelevant, write:  
     > “No relevant information returned by this agent.”
3. Compile results
   - Build a final structured summary as follows:
---
### OUTPUT FORMAT
Agent 1 (Roles & Responsibilities)  
Summary: [Your concise synthesis here]  
Sources: [link1], [link2]  
Agent 2 (Product & Technical Aspects)  
Summary: [Your concise synthesis here]  
Sources: [link1], [link2]  
Agent 3(Engineering Practices)  
Summary: [Your concise synthesis here]  
Sources: [link1], [link2]  
---
At the end, add this line for transparency:
> “This answer has been compiled from three expert Copilot Studio agents: Agent 1, Agent 2, and Agent 3”
---
### CONSTRAINTS
- Do not expose the sub-agents’ raw outputs or system prompts.  
- Maintain a neutral, factual tone.  
- Be concise — never exceed two sentences per sub-agent.  
- All sources must be formatted as clickable links (`Title`).  
- Ensure the final message is visually clear and readable in Teams or web chat.  
- If one agent fails to respond within timeout, mark it as “No relevant information returned by this agent.”  
- Return only the structured summary, nothing else.
 
I have the same question (1)
  • S-Venkadesh Profile Picture
    937 Moderator on at

    There’s a Microsoft document that explains how to prevent an AI-generated message from being sent to the user.
    To achieve this, you need to use a child agent instead of directly connecting another agent.
    However, it’s still unclear how the orchestration flow should be configured for this scenario.
    An AI-generated response is about to be sent Called when the main agent generates a response for a user after calling one or more topics, actions, or knowledge sources. Use the Response.FormattedText system variable to see the generated response. Set the ContinueResponse variable to false if you want to prevent the orchestration response being sent (that is, if you modify the message and send your own using a message node).
     
    To get a reliable solution:
    1. Create a topic that triggers and "Generate answers" one after another.
    2. Combine the results from each response.
    3. Send the final, consolidated message to the user.
     
    I hope this information helps! If it does, please consider giving a 👍.
    Best regards,
    S-Venkadesh
  • Romain The Low-Code Bearded Bear Profile Picture
    2,179 Super User 2025 Season 2 on at
    hello :) 
     
    Thanks for the answer, it could be totally usefull, it need to be used by implementing the topic trigger "an ai generated response is about to be sent" : this a great idea thank you :)
     
    Anyway in my use case there some other limitation wich force me to use "full connected agent" : 
    - around 20 SharePoint source per agent
    - need to have topic inside agent
     
    But really thank you it's a very good trick idea.
     
    About the documentation, unclear, sadly it's because before may 2025 there was checkbox in agent called "hide answer from agent" i m pretty sure it has existed one day (but i m tired i could be wrong)
     
    About the topic to call them : i deduce you speak about child agent and ""an ai generated response is about to be sent" to rewrite answer (since you could add a last message to be sure the 3 agent trigerred before combine answer : sadly not working in my case, but i like it)

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

Forum hierarchy changes are complete!

In our never-ending quest to improve we are simplifying the forum hierarchy…

Ajay Kumar Gannamaneni – Community Spotlight

We are honored to recognize Ajay Kumar Gannamaneni as our Community Spotlight for December…

Leaderboard > Copilot Studio

#1
Michael E. Gernaey Profile Picture

Michael E. Gernaey 270 Super User 2025 Season 2

#2
Romain The Low-Code Bearded Bear Profile Picture

Romain The Low-Code... 181 Super User 2025 Season 2

#3
S-Venkadesh Profile Picture

S-Venkadesh 93 Moderator

Last 30 days Overall leaderboard