AB-620: Predict your score.
Know what to study next.
See your predicted score, follow a study plan based on your answers, and revisit topics you're starting to forget. It all runs on your device.
387 practice questions
AI score prediction
100% offline
Download free
iPhone & iPad · Free to startOn your Mac? Scan to install
AB-620Microsoft AI Agent Builder
AB-620 · 30 days left ›
PREDICTED786 ±37✓ Exam Ready · 87% confidence
AB-620 Practice Questions & Exam Prep — Microsoft AI Agent Builder
Get exam-ready for AB-620 (Microsoft AI Agent Builder) on iPhone or iPad. Azure Mastery uses on-device AI to predict your readiness score across all three AB-620 domains, build a personalised study plan from your weak spots, and surface topics you're forgetting. Core study stays on-device and works offline; optional sync uses your private iCloud account.
The exam
What is the AB-620 exam?
AB-620 is the Microsoft AI Agent Builder credential — the role-based exam for designing and building integrated AI agent solutions in Microsoft Copilot Studio. AB-620 covers planning agent solutions, building and integrating agents in Copilot Studio (topics, generative answers, tools, MCP integrations, channels), and testing / managing agents in production (analytics, deployment lifecycle, governance).
AB-620 is hands-on and Copilot-Studio-aware. It covers:
Plan and configure agent solutions — use cases, environments, conversational design, integrations
Integrate and extend agents in Copilot Studio — trigger-phrase topics, generative orchestration, SharePoint and Dataverse knowledge sources, custom plugins via Power Platform connectors and MCP
Test and manage agents — test pane, analytics, channels, lifecycle, governance, security
Expect scenario questions that show you a Copilot Studio topic configuration or generative-answer setup and ask what to change.
Microsoft last updated the AB-620 study guide on 21 April 2026. Every question in Azure Mastery's AB-620 bank is mapped to the current outline. Read the official Microsoft study guide.
Questions40–60 questions, mixed formats
Duration100 minutes (120 min seat)
Pass score700 / 1000
CostUSD $165 (≈ £128 UK)
ValidityRenew annually (Associate)
FormatOnline or test centre
Study guide updated · 21 April 2026
AB-620 exam objectives
Three domains from Microsoft's current outline. Every domain below lists Microsoft's own skill groups verbatim from the official study guide, so you always know which domain you're being tested on and where your weak spots cluster.
AI & agent skill mapDataModelsAgentsResponsible AI
Plan and configure agent solutionsPublished weight 30–35%
135 exam-scoped practice questions in the app
Explore Plan topics
Plan an Agent Solution
Create and Monitor Agent Flows in Copilot Studio
Configure Topics
Covers audience, identity, channel, and governance choices made before any conversation logic gets built.
The identity-model decision gets skipped: a Teams-only agent and a public website agent need different auth models.
Learn how a topic calls a flow, a tool, or a custom-prompt node — that's the exam's real question shape.
Integrate and extend agents in Copilot StudioPublished weight 40–45%
170 exam-scoped practice questions in the app
Explore Integrate topics
Connect to Enterprise Knowledge Sources
Add Tools to Agents
Configure Multi-Agent Collaboration from Copilot Studio
Integrate Agents with Azure
The largest domain: grounding sources, connector types, and multi-agent delegation to Foundry, Copilot Studio, or Fabric agents.
Candidates can't tell a custom connector, an MCP tool, and computer-use automation apart for a given integration.
Practise wiring Azure AI Search or the Foundry model catalogue directly into a topic's generative-answer setup.
Test and manage agentsPublished weight 20–25%
82 exam-scoped practice questions in the app
Explore Manage topics
Evaluate Agent Performance
Implement Application Lifecycle Management (ALM) for Agents in Copilot Studio
Smallest domain but operational: building a test set, choosing an evaluation method, then packaging the agent for release.
Environment variables and connection references are what let a pipeline move an agent without manual copying — easy to forget.
Trace one agent through a full Power Platform pipeline from dev to production before exam day.
Common traps
Where AB-620 candidates slip
Five recurring misconceptions that trip up otherwise well-prepared AB-620 candidates, grounded in the current skills outline.
Connector vs. MCP tool vs. agent flow
Candidates can't tell a connector, an MCP tool, and an agent flow apart when a scenario needs one of the three.
A stable API points to a connector, a dynamically discovered schema points to an MCP tool, and multi-step orchestration points to an agent flow.
Look for the named constraint in the scenario — stability, discovery, or approvals — it picks one of the three for you.
Topics handle conversation, flows handle the backend
Candidates treat an agent flow as just a topic with extra steps, missing which layer owns a piece of logic.
A topic decides what the agent says next; an agent flow decides what happens in the systems behind it.
Ask whether the described behaviour is what the agent says or what happens behind the scenes before answering.
A2A protocol vs. an embedded sub-agent
Candidates read "integrating a Foundry agent" without noticing whether it's a separate A2A agent or an embedded sub-agent.
A child agent receives the parent agent's context and uses its own tools and knowledge. A connected A2A agent remains a separately managed agent, with context hand-off controlled by the connection.
Check the endpoint, context setting, and authentication requirements instead of assuming permissions are inherited.
Groundedness rules aren't optional
Candidates assume a correct-sounding answer is enough, but Copilot Studio can withhold a knowledge-source response when ungrounded responses are disabled and the answer lacks an in-text citation.
Groundedness evaluation checks whether an answer is supported by the supplied context; citation behaviour is a separate product setting and response requirement.
For a "confidently wrong agent" scenario, inspect the knowledge source, orchestration, citation, and ungrounded-response settings before blaming the model.
Identity choice depends on audience, not habit
Candidates default to one identity model out of habit instead of matching it to who or what is calling the agent.
User-delegated tokens, service principals, and managed identities each map to a different kind of caller.
Ask whether a known Entra user or an unauthenticated external caller is making the request before choosing an identity model.
Designed for AB-620
How Azure Mastery helps you pass AB-620
Exam-specific practice
387 AB-620 practice questions aligned to the published skills outline.
Coverage leans hardest on integrating and extending agents, the domain Microsoft weights up to 45%.
Practise picking a connector, an MCP tool, or an agent flow for a stated integration need.
Predicted score
Exam IQ forecasts your AB-620 score on-device after roughly 30 questions, confidence range attached.
It flags whether integration breadth or agent testing and ALM is dragging that number down.
See which of the three domains needs the most work before you book the exam.
Adaptive study plan
Your plan leans hardest on integrating and extending agents, the domain Microsoft weights up to 45%.
Miss a grounding-source or identity-model question and the next session surfaces that domain first.
Stay solid on a topic across three sessions and the plan shifts to your next weak spot.
Knowledge decay
Forget the connector-vs-MCP-tool distinction early on and later integration questions get harder too.
A decay flag on the Today screen tells you which topic to revisit before it fades.
Focus Weak Spots sessions bring those fading topics back into rotation for you.
Exam rehearsal
Practise holding focus through a full 100-minute session, matching AB-620's real length.
Original questions in each session follow the published domain weighting, not a fixed script.
Microsoft sets the live exam's interface and exact question order, not Azure Mastery.
Answer Coach
Untangles AB-620's near-identical pairs: connector vs MCP tool vs agent flow, and A2A agents vs embedded sub-agents.
Get a question wrong and Answer Coach shows the exact scenario clue that should have picked a different tool or identity model.
An optional on-device rewrite is checked against the written explanation before you see it.
Aura guidance
Aura adjusts its suggestions as your first week of AB-620 practice takes shape.
Every session closes with a quick recap: what improved, what needs another look, and what's next.
Private by design
Your AB-620 answer history, readiness gauge, and Answer Coach notes stay on your device by default.
There's no Azure Mastery account or server that ever sees your study data.
Switch on sync and only your own private iCloud account ever carries the data.
6-week study plan
Suggested AB-620 study plan
Plan an agent, build its conversations and integrations, then test and manage it. Use this six-week route at a pace that suits your Copilot Studio experience.
Plan the agent
Identify use cases and compare custom agents with templates.
Plan Dataverse environments.
Choose channels such as Teams, Microsoft 365, web or voice.
Design conversations
Define entry points and fallback responses.
Plan escalation to a person.
Review DLP, sharing and environment security roles.
Build topics and orchestration
Create topics and trigger phrases.
Use conversation nodes, variables and branching.
Explore generative orchestration for dynamic responses.
Connect knowledge and tools
Add knowledge from SharePoint, Dataverse, websites, files or Azure AI Search.
Connect Power Automate, MCP and AI Builder tools.
Configure Microsoft Entra and OAuth authentication.
Test and manage
Inspect variables, trace topics and debug orchestration.
Publish to channels and monitor analytics.
Manage versions, solutions, environments and governance policies.
Review and rehearse
Use Focus Weak Spots to revisit weaker topics.
Complete two timed, 100-minute Exam Simulator sessions.
Review mistakes and repeat the areas that need practice.
Inside the app
Nine interactive practice formats, on iPhone
Azure Mastery has nine interactive formats for exam practice. The examples below show original AB-620 practice questions in Azure Mastery on iPhone and iPad. The exam simulator uses timed sessions with the published domain weights. Microsoft's live interface and question mix may differ.
Which primitive in MCP represents an executable capability that the agent can invoke with arguments?
Prompt
Transport
Resource
Tool
Multiple choice
An original AB-620 practice question with one correct answer. The app explains every option after you answer.
Exam-specific sample
Which Microsoft Fabric item types can be attached as data sources when building a Fabric Data Agent? Select all that apply.
Power BI semantic model
Lakehouse
Warehouse
Azure DevOps repository
Multi-select
An original AB-620 multi-select question. Select all the correct options to earn the mark.
All-or-nothing
Order the steps to build and publish a basic Copilot Studio agent.
⋮⋮1Create the agent and give it a name, description, and instructions
⋮⋮2Add knowledge sources and topics to the agent
⋮⋮3Test the agent in the Test chat pane
⋮⋮4Publish the agent to the chosen channel
Drag-and-drop
An original AB-620 ordering question you can answer by touch on iPhone and iPad.
Interactive item
In a Copilot Studio agent, which element should you add so the agent can answer from your indexed enterprise content?
Hotspot
An original AB-620 question with a visual prompt, designed for touch on iPhone and iPad.
Tap target
Contoso — Building an enterprise support agent in Copilot Studio Contoso wants an internal support agent in Copilot Studio that answers employee questions from a large SharePoint and ServiceNow knowledge base, performs ticket…
1How should the agent ground its generative answers on the enterprise knowledge?
2Where should the agent be deployed so employees can use it where they already…
3How should the agent automate a repeatable back-end task during a conversation?
4How should the agent take actions in ServiceNow?
Case studies
An original AB-620 case study with several questions about the same requirements and environment.
Multi-question
✕Your answer: Prompt
✨ Answer Coach:A prompt in MCP is a reusable instruction template the host can apply, not a function the agent executes with parameters, so it does not model a runnable capability.
— grounded in authored certification guidance
Answer Coach
Answer Coach explains why the answer is correct and, where available, why each option is right or wrong. It helps you understand a mistake and remember the distinction. Read explanations after each question or at the end of a practice test. On supported devices, optional AI can reword a note on your device after checking it against the written guidance.
App exclusive
Frequently asked
AB-620 FAQs
How much does the AB-620 exam cost?
The AB-620 voucher costs USD $165 in the United States (around £128 in the UK; pricing varies by region). Watch for occasional free-voucher promotions around Microsoft Build or Microsoft Ignite — Microsoft doesn't run them year-round. Renewal itself is free.
Does the AB-620 certification expire?
Yes — AB-620 expires annually like every Microsoft Associate certification. Renewal is a free, short online assessment on Microsoft Learn, available in the six-month window before expiry and scoped to recent skills-outline updates. (Fundamentals certifications such as AZ-900 are different — those don't expire.)
What is the AB-620 retake policy if I fail?
Retake timing scales up with each failed attempt, and every attempt needs its own voucher.
First retake: after 24 hours.
Second and third retakes: a 14-day wait each.
Cap: five attempts per rolling 12 months.
What does the AB-620 exam format look like?
AB-620 runs as a 100-minute session (120 minutes with seat time) of 40–60 questions, taken online with a proctor or at a test centre, and scored out of 1000 with a 700 pass mark. Azure Mastery's bank includes single-answer, multi-select, yes/no, and matching items, so it's not straight recall.
How long should I study for AB-620?
Four to eight weeks if you already build with Copilot Studio or the wider Power Platform; longer if Power Fx, Dataverse, MCP, or agent lifecycle management are unfamiliar. The exam leans heavily on the Integrate and extend domain (40–45%), so prioritise hands-on time there.
AB-620 vs AI-103 — different agent roles?
Different platforms for different builders. AB-620 covers Microsoft Copilot Studio, a low-code environment for topics, generative answers, and Power Platform integrations. AI-103 covers Microsoft Foundry, a code-first platform for Python-based agents and orchestration.
AB-620 vs AI-901 — which should I take first?
Take AI-901 first if AI concepts are genuinely new to you — it's the Fundamentals exam and assumes no build experience. AB-620 is role-based and assumes you can already navigate Copilot Studio, Dataverse, and Power Platform environments.
Do I need a Copilot Studio licence to prepare for AB-620?
Not to use Azure Mastery, but hands-on practice matters more for AB-620 than for most exams. Microsoft offers trial access to Copilot Studio — check the current terms on Microsoft Learn — and building topics, agent flows, and knowledge-source connections there pairs well with Azure Mastery's scenario questions, rather than relying on question practice alone.
Is Azure Mastery free for AB-620 prep?
The app is free to download with a free allowance of AB-620 questions. The full 387-question AB-620 bank unlocks with a one-time exam-pack purchase, or a subscription or lifetime upgrade unlocks every exam in the catalogue.
Does AB-620 practice work offline?
Yes. The whole study loop, from answering questions to the readiness prediction, runs on the phone, so a flight or a signal-free commute is still study time. There is no account to create, and your progress stays on the device unless you switch on private iCloud sync.
Free study guides
Free guides that pair with AB-620
Where AB-620 fits
Certification paths that include AB-620
AB-620 is the Microsoft AI Agent Builder credential — the role-based cert for Copilot Studio builders. It follows AI-901 as the current fundamentals route and complements AI-103 for candidates who span both low-code (Copilot Studio) and code-first (Microsoft Foundry) agent platforms. Tap any linked exam below to see its dedicated study app page.
AB-620 sits at the AI Agent Builder Associate tier. AI-901 is the current fundamentals foundation; AI-103 is the natural complement for candidates who also work with code-first agents in Microsoft Foundry.
Ready to pass AB-620?
Download Azure Mastery free. 387 AB-620 practice questions across all three domains, AI score prediction, full-length exam simulator, adaptive study plan. iPhone & iPad.
