6.2 Configure Agent Response Formatting

Microsoft Certified: AI Agent Builder Associate AB-620

Track your progress as you read

Finished sections get a check mark.

1 Introduction & Exam Overview

1.1 Current Exam Facts 1.2 Candidate Profile & Prerequisites 1.3 Skills Map & Study Strategy

2 Plan Integration, Identity & Channels

2.1 Plan Integration with Enterprise Systems 2.2 Plan Identity Strategy 2.3 Plan Channels and Deployment

3 Responsible AI, Security & Reuse

3.1 Plan Responsible AI Strategy 3.2 Security and Governance Considerations 3.3 Plan Reusable Agent Components 3.4 Design Agents for Internal or External Audiences

4 Agent Flows Fundamentals

4.1 Create an Agent Flow 4.2 Human-in-the-Loop Agent Flows 4.3 Configure Actions and Connectors in Flows

5 Agent Flow Monitoring & Resilience

5.1 Monitor Agent Flows 5.2 Input and Output Parameters 5.3 Error Handling in Agent Flows

6 Topics Configuration Core

6.1 Add Agent Flows to a Topic 6.2 Configure Agent Response Formatting 6.3 Add Tools to a Topic

7 Advanced Topic Responses

7.1 Custom Prompts for Advanced Responses 7.2 Custom Knowledge Sources for Advanced Responses 7.3 API and Send HTTP Request Responses

8 Generative Answers, Cards & Variables

8.1 Configure the Generative Answers Node 8.2 Configure Adaptive Cards 8.3 Manage Variables

9 Enterprise Knowledge Sources

9.1 Connect to Copilot Connectors 9.2 Connect to Power Platform Connectors 9.3 Connect to Azure AI Search

10 Add Tools to Agents

10.1 Configure and Monitor Computer Use 10.2 Configure MCP Tools 10.3 Tools via Existing Custom Connectors 10.4 Add REST APIs to an Agent

11 Multi-Agent Collaboration

11.1 Design Multi-Agent Solutions 11.2 Integrate a Foundry Agent 11.3 Integrate an Existing Copilot Studio Agent

12 Fabric Data Agents & A2A

12.1 Integrate a Fabric Data Agent 12.2 Multi-Agent Solutions with A2A Protocol

13 Integrate Agents with Azure

13.1 Generative Answers with Azure AI Search & Foundry 13.2 Foundry Model Catalog for Custom Prompts 13.3 Monitor Agents with Application Insights

14 Evaluate Agent Performance

14.1 Create a Test Set 14.2 Choose an Evaluation Method 14.3 Review Test Results

15 Application Lifecycle Management

15.1 Create a Solution 15.2 Add Existing Agents to a Solution 15.3 Create and Use Environment Variables 15.4 Power Platform Pipelines for Agents

6.2 Configure Agent Response Formatting

Key Takeaways Explain This Topic AI

Message nodes deliver text with basic formatting, variables, images, videos, basic cards, Adaptive Cards, quick replies, and optional message variations.

Channel constraints matter: some channels strip rich formatting, limit quick replies, or render Adaptive Cards differently—design for the target channel set.

Quick replies suggest next steps but do not force a choice; Question nodes constrain input when a required option is mandatory.

With generative orchestration, tools and topics can return outputs for AI-authored responses, or you can send specific templated text or Adaptive Cards on completion.

Speech overrides and SSML customize voice channels separately from the visual text shown in chat.

Last updated: August 2026

6.2 Configure Agent Response Formatting

Quick Answer: Use Message nodes for authored text (bold/italic/lists), variables , media, basic cards , Adaptive Cards , quick replies , and variations . Respect channel limits . Under generative orchestration, choose whether tool/topic completion should AI-write , send a specific response , or send an Adaptive Card . Use speech overrides/SSML when voice channels matter.

AB-620 Domain 1 includes Configure agent response formatting because an agent that returns correct data in an unreadable blob still fails users—and exam scenarios often ask which formatting construct fits a channel or UX goal.

Message nodes: the primary formatting surface

Within a topic, the Message node sends content from the agent to the user. The same formatting capabilities are available on Question nodes when you prompt for input.

Text formatting toolbar

The message node supports lightweight formatting without custom code:

Capability Typical use Channel note

Bold / italic Emphasize ticket numbers, deadlines, warnings Usually fine in Teams and web chat; some channels strip styles

Bulleted / numbered lists Steps, policy conditions, multi-item results Prefer short lines for mobile
Line breaks Separate greeting, body, call to action Avoid huge walls of text
Variables {x} Insert topic, global, or system variables at runtime Resolve at send time

Personalization pattern: greet with System.User.DisplayName , bold the entitlement total, then list remaining leave types. The information is unchanged; scannability improves.

Message variations

Message variations let you define multiple phrasings on one node. Each run, the agent randomly picks one variation. Use variations for greetings and repeated confirmations so the agent feels less robotic— without changing policy substance. If variation A mentions a deadline, every variation should mention the same deadline.

Images and videos

From the message Add menu:

Image: publicly reachable image URL; optional title.

Video: public MP4 URL or YouTube URL; optional title, subtitle, image, text, buttons.

Assets must be publicly accessible . Internal-only SharePoint files without appropriate sharing, or authenticated intranet URLs the channel cannot fetch, will not render.

When a node has multiple media cards:

Carousel (default): one card at a time.

List: all cards vertically—better for side-by-side comparison.

Cards and structured visual responses
Basic card

A basic card is a general-purpose layout for text, image, and interactive buttons (same action ideas as quick replies). Use it for simple promotions or single-item highlights when you do not need Adaptive Card schema power.

Adaptive Cards in message nodes

Adaptive Cards are platform-agnostic JSON layouts. On a Message node:

Add → Adaptive card .

Open Edit adaptive card (designer or card payload editor).

Author elements or paste JSON; use Power Fx where dynamic values must bind into the card.

Save and preview on the node.

Use Adaptive Cards when structured fields matter: order lines, leave balances, multi-column summaries, status chips. Informational cards on Message nodes display ; collecting multi-field input often uses Ask with Adaptive Card (interactive) covered more deeply with adaptive-card skills later in the guide.

Channel reality check: Teams, web chat, and Microsoft 365 Copilot surfaces may enforce different Adaptive Card rules (for example validation guidelines for rich responses in Microsoft 365 experiences). Always test the target channel , not only the Copilot Studio test canvas.

Quick replies vs forced choices

Quick replies are suggestion chips under a message:

Quick reply type Behavior
Send a message (default) Sends visible text back to the agent
Open URL Opens https:// link
Make a call PhoneNumber: plus number
Send a hidden message Sends payload without showing it in history

Important limits:

Some channels do not support quick replies (buttons simply missing).

Some channels cap how many chips appear.

Users can ignore quick replies and type freely.

When the conversation must not proceed without a recognized option, use a multiple-choice Question node instead of quick replies. Exam wording often hinges on guide (quick reply) vs constrain (question).

Generative orchestration and response completion

Response formatting is not only Message nodes. When tools (including agent flows) complete under generative orchestration, Completion settings control what the user sees:

After running option Effect
Don't respond (default) Agent incorporates tool output into its broader generative reply
Write the response with generative AI Model crafts a contextual answer from tool outputs
Send specific response You template exact text with output variables / Power Fx
Send an adaptive card Rich interactive/visual result bound to outputs

Topics can also return output parameters and let the orchestrator summarize, rather than always ending every topic with a hard-coded Message node. That pattern increases flexibility but reduces word-for-word control—choose based on compliance.

Responsible UX tip: Microsoft recommends telling users that some conversation (for example AI-generated clarification questions) may be generated. A common place is an extra line in the Conversation Start system topic.

Channel constraints matrix (memorize patterns)

Surface Strengths Watch-outs
Microsoft Teams Cards, many rich features Card schema version and action support nuances

Custom website / demo site Full message node features in many cases Host media publicly; brand CSS is not your Adaptive Card engine

Voice / Omnichannel voice Spoken experience Configure speech message override ; optional SSML (break, emphasis, prosody, audio)

Channels without quick replies Still get text Do not rely on chips as the only path

External / multi-agent paths Orchestration power Some multi-agent handoffs have reported card rendering limits—verify

Speech override pattern: message text can drive both text and voice by default; expand the selector to Speech and author a spoken-friendly line, then add SSML tags when pitch, rate, or pauses matter.

Scenario walkthroughs
Scenario A — HR leave balance

After a Dataverse tool returns balances, prefer Send specific response or a Message node:

Bold remaining vacation days.

Numbered list of upcoming blackout dates from a variable.

Quick replies: “Request leave,” “View policy,” “Talk to HR.”

If the balance is multi-field (vacation, sick, floating), an Adaptive Card with labeled columns beats a paragraph.

Scenario B — Store hours with media

Message node text for hours + Image of the storefront + Basic card button “Get directions” (Open URL). Test on the customer’s actual web channel for image hotlink rules.

Scenario C — Generative order status

Agent-level REST tool completion set to Write the response with generative AI for casual status chat. For regulated refund amounts, switch to Send specific response so currency formatting cannot drift.

Scenario D — Voice self-service

Visual channel shows a short confirmation code in bold; Speech override spells the code slowly with SSML breaks so callers can write it down.

Authoring quality checklist

Lead with the outcome (“Your ticket INC0123 is open”) before background.

Insert variables instead of hard-coding user-specific values.

Keep variations semantically identical.

Host media on durable public URLs (or approved CDN/storage).

Design fallback plain text when cards or chips may not render.

Match completion mode to risk: generative prose for low-risk Q&A; templated or Adaptive Card for transactional certainty.

Test every production channel , including mobile widths.

For generative agents, ensure topic/tool descriptions stay accurate so the right formatting path is even selected.

Exam traps

Treating quick replies as mandatory multiple choice.

Assuming Adaptive Cards render identically on every channel without testing.

Putting secrets or internal-only image URLs into Message nodes.

Forgetting speech overrides on voice deployments.

Using only generative free text for legally fixed disclosures that must be word-perfect.

Response formatting is product UX engineering inside Copilot Studio. On AB-620, show you can pick the right node, media, card, completion mode, and channel-safe design—not only that you can “send a string.”