AB-100 practice questions
Microsoft · AB-100 · 300 questions
Original practice questions for the Agentic AI Business Solutions Architect exam, covering every domain of the official exam objectives.
This course contains the use of artificial intelligence.
About the AB-100 exam
- Questions
- not published
- Passing score
- 700 or greater
- Format
- not published (question count and duration are not on the study guide or exam page)
Schedule this exam The certification this earns
Exam details published by the vendor, checked 7 November 2026. Vendors change fees and formats without notice — confirm on the vendor's own page before you book.
Practice Quizzes
Test your knowledge with standard 20-question practice sets.
Quiz 1
Quiz 2
Quiz 3
Quiz 4
Quiz 5
Quiz 6
Quiz 7
Quiz 8
Quiz 9
Quiz 10
Quiz 11
Quiz 12
Quiz 13
Quiz 14
Quiz 15
Browse by Domain
Study specific topics at your own pace.
Plan AI-powered business solutions (35-40%) · 122 questions
- A regional grocery chain wants an agent that places a purchase order when a store's on-hand count for a staple item falls below the reorder point, and messages the manager only if the supplier rejects that order. Which use of an agent matches this requirement?
- A plant copilot quotes a retired torque value because last year's work instruction is still in its grounding library and the current controlled document is missing from that library. What should the architect correct first?
- A retailer claims an agent project is profitable because the Copilot Studio license costs less than the wages of the clerks it might deflect. Maker time, model consumption, support, and training are left out. What should the architect add before accepting that return?
- Claims adjusters at an insurer already work in Outlook and Teams. They need answers drawn from their own mail and from the claim file in SharePoint, inside the Copilot they already open. The agent has no public website and no separate sign-in. What should the architect choose?
- Several university agents need the student handbook. The current file sits in a private team most agents cannot read, and a stale public PDF is what their knowledge sources index. What should the architect do with the data?
- A logistics sponsor wants a custom model trained on last month's yard photos so an agent can say which dock is free. The yard system already exposes the live dock status, and the status changes by the minute. What should the architect decide?
- A public agency's sponsor wants dozens of agents in production next quarter. No one has ranked the use cases, and no proof of concept has been run. The agency already has an Azure landing zone. What should the architect do first?
- A bank sends every request, from a one-line FAQ rewrite to a multi-step credit-memo argument, to its largest model deployment. Cost is climbing and simple requests are no faster. What should the architect introduce?
- An airline wants an agent that retickets disrupted passengers under a published rule and messages a lead only when the fare jump exceeds a written cap. Which use of an agent matches this requirement?
- A hotel group wants an agent that forecasts weekend occupancy for the revenue manager's Monday meeting and then stops. It must not change room rates. Which use of an agent matches this requirement?
- A city transit agency wants an agent that ranks route options for a planner, who must accept or reject that ranking before any schedule is changed. Which use of an agent matches this requirement?
- A pharmacy chain wants an agent that explains which stores are stocking out of a medicine and why, and that must not place a purchase order. Which use of an agent matches this requirement?
- Sellers at a professional-services firm want meeting prep that combines the customer record in Dynamics 365 with their recent mail. They do not have a unique sales process. What should the architect propose first?
- Each department at a retailer builds agents with its own data rules, and leadership cannot tell which agents exist or whether any of them paid off. There is no shared intake. What should the AI Center of Excellence establish first?
- A farm cooperative wants an agent that recommends which fields to irrigate tonight. A grower must confirm the list before any valve opens. Which use of an agent matches this requirement?
- A harbor terminal wants an agent that summarizes crane idle time for the operations director and must not dispatch labor. Which use of an agent matches this requirement?
- A bank's product agent quotes a brochure that is accurate for a product this customer is not allowed to buy. Products the customer can buy sit in another library the agent does not search. What should the architect correct?
- A hospital discharge agent mixes two patients in one answer. The grounding folder holds scans with no encounter identifier, so a file cannot be tied to one visit. A teammate wants a new model. What should the architect correct first?
- An insurer's claim agent cites a policy rider that was withdrawn last quarter. That withdrawn file is still in the grounding index, beside the current rider. What should the architect correct first?
- A utility's outage agent answers on quiet days and fails on a storm day. The live outage-status feed is not in the grounding set, so the agent can read only last month's static map. What should the architect correct?
- An airline keeps its crew-legality rules in a PDF on one planner's laptop. The crew-scheduling agent and the payroll agent both need those same rules. What should the architect do?
- A city transit agency pasted its fare table into the instructions of three fare agents. When a fare changes, only one of those instructions is updated. What should the architect do?
- A hospital keeps allergy lists in a file that only one departmental bot can open. The chart agent and the pharmacy agent both need those lists, and access must stay governed. What should the architect do?
- A manufacturer copied torque specifications into a chat log that only the night-shift agent reads. Day-shift agents need the same specifications, and the controlled-document library is the approved source. What should the architect do?
- A retailer copilot answers size questions with blanks. The size charts are images, and the grounding index stored those pages with no extracted text. What should the architect correct?
- A university admissions agent answers from two current sources that disagree. A campus blog says a program is open to first-year students, and the official catalog says it is not. Which grounding change should the architect make?
- A bank's platform team wants a new landing zone that contains only AI workloads. The bank already operates a landing zone with an approved governance baseline. What should the architect do?
- An insurer keeps its fraud glossary inside one agent's prompt. The claims agent and the special-investigations agent both need those terms, and access to the glossary must be reviewed in one place. What should the architect do?
- A consultancy's proof of concept met its success criteria. The sponsor wants to build the rest of the backlog immediately and skip the roadmap. What should the architect do with the proof?
- A utility's executives selected a model vendor before anyone named a use case, a data owner, or a responsible AI review. What should the architect put in place first?
- A public agency wants an agent grounded on case files, but the data owners will not name who is allowed to use those files for grounding. The sponsor asks for a larger model instead. What should the architect treat as the blocker?
- A hotel group wants its first proof of concept to answer guest complaints on the public website immediately. No use case has been ranked or proven yet. How should the Cloud Adoption Framework plan start?
- An insurer's draft strategy says no agent may take an action, including an agent that only reads a policy to answer a question. What should the architect change?
- A university's AI strategy names a model to acquire and does not say what any use case should improve. What should the architect do with that document?
- A pharmacy chain's AI strategy is a list of product and tool names and does not say what should change in the store. What should the architect require before any build is funded?
- A logistics firm wants an agent that reads order status. An order API already returns that status to other applications. A proposal funds a new orchestration layer before any agent work. What should the build strategy do?
- An airline sponsor wants one agent to schedule crews, refund passengers, and write maintenance logs. What should the build strategy do with that request?
- Support, store operations, and finance at a retailer each want an agent this quarter. The architect must set the build order before any team starts. What should decide that sequence?
- A logistics firm uses a Microsoft Foundry agent to score delay risk from internal shipment data, and a Copilot Studio agent to tell the customer the new arrival window. A sponsor wants the score folded into the customer chat. What should the architect keep?
- Citizens of a public agency may ask only about their own case, while caseworkers may ask about any case in the unit. The two audiences have different data rights. A designer wants one prompt for both. What should the architect require?
- Store associates at a retailer ask policy questions in the Microsoft 365 apps they already use, against their own work files. Shoppers on the public returns website have no work account and must start a return there. Which build should the architect choose?
- A maintenance agent and a safety agent both write the same work-order fields and overwrite each other's updates. What should the design change?
- A bank's customer-service agent must hand a suspected fraud case to a fraud agent. The customer-service agent must not receive the full transaction archive. What should the design set?
- A hospital built one agent that answers scheduling, benefits, and clinical questions. It now gives clinical answers it is not authorized to give. What should the plan change?
- University IT already answers published how-to questions with a prebuilt help agent. The same team must also grant a badge-access request, which that prebuilt agent cannot do. What should the architect build?
- A university admissions agent and a financial-aid agent both answer grant questions, each from its own copy of the aid policy. The copies now disagree, and students hear two awards for the same question. What should the architect change?
- Hospital billing wants a plain-language collections explanation from invoice data Dynamics 365 Finance already holds. A proposal builds an agent that retypes each invoice into a chat before any finance copilot is tried. What should the architect choose first?
- Insurance sellers already prepare for meetings with the prebuilt sales copilot, using the customer record and their mail. Another team wants a custom prep agent with its own brief. Sellers would then prepare in two places. What should the architect choose?
- A retail service desk wants help with password resets and order-status questions. A prebuilt service copilot already matches that work. The sponsor wants to start by training a custom model on old chats. What should the architect propose?
- Warehouse staff ask on-hand questions that a supply-chain copilot already answers from the live warehouse system. The copilot does not know the local rule for which trailer may wait in the yard. What should the architect do?
- A utility's prebuilt field copilot already summarizes the work order for the crew. It does not know the local rule that says when a crew must stop and call the control room. What should the architect add?
- A bank wants Copilot Studio to answer questions about deposit products, and the same agent must not approve credit. When should the architect write that limit?
- An airline wants a Copilot Studio agent that can book a special meal and must not change which aircraft is assigned to the flight. Where should that boundary sit?
- A public agency will ground an agent on citizen case files. Policy forbids using those files to tune a model. What must the build plan state?
- A hospital wants Microsoft Foundry to host a model that suggests clinical codes from chart text. That model must not be the system that files the chart. Which constraint fits?
- A retailer wants receipt fields taken from a photo and a conversation with the shopper about the return. Which split matches the constraint?
- A retailer's returns agent is asked for a refund the current policy page does not mention. Which behavior should the architect require?
- A pharmacy agent is asked whether two drugs interact. Which behavior should the architect require for that answer?
- A consultancy's client requires every agent answer to cite a source in the engagement file. Which design does that constraint remove?
- A bank wants product questions answered from its approved FAQ library, and it wants complaint outcomes kept off that library. Which knowledge design fits?
- A university agent may answer course-catalog questions with generative knowledge. Degree-audit exceptions must not be answered that way. Which plan fits?
- A hotel agent is asked whether a specific room has a crib. Properties differ, and a typical hotel is not a safe guess. Which source should the architect require?
- City clerks answer residents on a public website. Extending each clerk's Microsoft 365 Copilot would use that clerk's own files. Which build strategy fits?
- Airline crew already work in Teams and need answers drawn from their own roster mail. Which build strategy fits?
- Partners at a consultancy draft proposals in Word and want clause suggestions drawn from their own files. Which build strategy fits?
- Hospital administrators work in Outlook and need policy answers drawn from their own mailboxes. Which build strategy fits?
- Factory supervisors work on a shop-floor kiosk and do not have a Microsoft 365 license. They need an agent for shift procedures. Which build strategy fits?
- A logistics agent should say where a shipment is by using the shipment knowledge connection, and it should not invent an excuse when a delay has no source. Which design fits?
- Utility inspectors need one defect type named on pole photos. The vision service still cannot label that type after the team supplied examples, and a stable set of labeled photos already exists. Which approach fits?
- A hospital wants dictated clinician notes turned into readable text before they are filed. A Foundry Tools speech skill already performs that cleanup. A vendor proposes training a custom model on the same audio. What should the architect choose?
- A bank sponsor asks for a custom model because the policy PDFs use an unusual brand font. The only job staff need is for the copilot to read the policy text they ask about. What should the architect decide?
- A retailer wants an agent to answer whether a SKU is on sale. The price API already returns that flag. What should the architect decide?
- A retailer wants an agent on the point-of-sale register. The register is not a Microsoft 365 app, and the associate's Copilot cannot run there. Which build strategy fits?
- An insurer must classify fraud notes into a stable set of labeled types. Grounding and worked examples still leave the foundation model on the wrong type. Which choice fits?
- At a government agency, staff paste prompts from personal notes into production citizen agents, and two agents now instruct citizens differently. What should the architect establish?
- Clinicians at a hospital forward a useful prompt by email, and several production agents now contain different pastes of that message. What should the architect do with the prompt?
- A university prompt that is already in the approved library tells the agent to invent a policy when it is unsure. Students have received policies the registrar did not write. What should the architect revise?
- A retailer's prompt library already holds many prompts and has no named owner. Several teams want to add more prompt text this week. What should happen first?
- Two teams at a bank each put a prompt for the same fee-waiver decision into the prompt library, and the two prompts contradict. What should the library do?
- A retailer must classify each return into one of twenty known reason codes. Volume is high, the code list stays fixed for the season, and nobody needs an open conversation. Which model choice fits?
- A bank wants a small language model to be the only reviewer of suspicious-activity narratives that depend on account history, prior cases, and the written policy. What should the architect decide?
- A hospital wants a small language model to give open-ended care advice when a nurse asks what to do next and no fixed protocol applies. What should the architect keep for that judgment?
- A logistics yard receives a short gate message that contains a container id. That same step must not bargain with the customer about fees. Which design fits?
- A regional airline must tag every flight delay note into a fixed set of reason codes, across that day's departures. Which model choice fits?
- An insurer's coverage prompts say only to be helpful, and the agent states benefits the policy text does not contain. What should the prompt guidelines require?
- A university policy bot is prompted to write a long essay for every question, and students cannot find the rule they asked about. What should the prompt guidelines set for the answer?
- A bank's AI Center of Excellence keeps its agent standards on a slide, but intake still accepts proposals that ignore those standards. What should the CoE change?
- A hospital keeps approving another agent pilot while nobody has checked whether the departments have the skills to run an agent. What should come before the next pilot?
- A hotel guest agent runs on a single prompt line with no stated role, and it has offered room upgrades the property does not sell. What should the guidelines add?
- A bank's prompts tell the agent that it is the compliance officer and that it issues the decision on the case. What should the prompt guidelines bar?
- A manufacturer's prompts paste the entire service manual into every call, and that manual is revised several times a year. What should the prompt guidelines say?
- An airline's agent pilots succeed, and the next team starts from a blank page. The AI Center of Excellence sends congratulations and stops. What should it do with a successful pilot?
- A distributor's sales copilot states a quote that the finance copilot will not invoice, and each copilot can see only its own app. What should the architect design?
- A public agency's AI Center of Excellence reports how many agents it published and how many workshops it held. Sponsors funded a cut in the permit backlog, and that result is missing. What should the CoE change?
- A utility's AI Center of Excellence has no intake. During storm week the loudest request gets an agent, and planned safety work waits. What should the CoE create?
- A hospital buyer agent can see a purchase request in Supply Chain and cannot see that Finance has placed that vendor on hold. What should the design make visible to that same agent?
- A university finance copilot tells a student one balance, and the customer service copilot tells an advisor a different balance for that same student. What should the design share across the two apps?
- Every brand at a retailer hires its own AI lead, and those leads ignore the Center of Excellence. Leadership asks what mandate the CoE needs. Which response fits?
- A hotel sponsor ranks agent projects on license cost and call deflection, and the criteria ignore the time managers will spend on change management. What should be added before the projects are compared?
- A city transit agency ranks agent projects by the license price alone and treats that price as the entire cost. Maker time, model consumption, support, and training are left out of the criteria. What should the architect change?
- An insurer changes a policy in the sales app and later handles a claim on that policy in the service app. The two copilots were built as if the work were unrelated. What should the architect change?
- A manufacturer opens a field case in Customer Service, then orders the part through a Supply Chain copilot that cannot see that case. What should the architect design?
- A bank ranks agent proposals by model consumption alone. Call deflection, the benefit the sponsor funded, is missing from the criteria. What should the criteria add?
- A consultancy's return criteria rank agent projects by the size of the model each one would call, so the largest model wins the funding. The process each project would improve is not in the score. What should the criteria score instead?
- A retailer treats each deflected chat as a full saving. Many of those chats come back the next day as store visits, and the analysis does not count the returns. What should the analysis count?
- A hospital prior-authorization agent reduces clerk time and adds clinician review time. The analysis reports the clerk hours saved and leaves the new clinician time out. What should the analysis do?
- An airline's rebooking agent shortens calls. When it selects the expensive fare it also creates a compensation cost, and the current analysis leaves that cost out. What should the analysis include?
- A utility's storm agent shortens outage calls and will not dispatch a crew until a new on-call approver agrees. The analysis lists the saved call time and omits the approver. What should the analysis include?
- A university admissions agent answers applicant questions after the office closes. The draft analysis claims the benefit is a cut in admissions headcount, and no role is being removed. What should the analysis treat as the benefit?
- A pharmacy chain wants to rank several agent ideas, but the criteria include no baseline. Nobody has recorded what the manual dispensing follow-up costs today. What should the architect define before any idea is ranked?
- A manufacturer's maintenance agent avoids one class of unplanned downtime on a single line. The draft analysis claims company-wide savings across plants and lines the agent does not watch. What should the analysis use as the benefit?
- A public agency must issue a statutory notice that no commercial agent produces. A product the agency can configure already covers the surrounding citizen FAQ. What should the architect build?
- A hospital wants a scheduling copilot that already books visits. No product on the shortlist enforces the hospital's local consent rule for minors. How should the architect close the gap?
- A logistics team wants a tracking assistant for shipments. The carrier portal they already license includes that assistant, and it can be configured for their account. A proposal would rebuild the same assistant from scratch. What should the architect decide?
- A retailer watched a vendor demo whose script matched the sales slideshow. The demo did not apply the retailer's returns policy, which is the process the sponsor needs. What should the architect decide?
- A bank's prebuilt service copilot already handles card disputes and is missing only the current fee table. Leadership asks about funding a ground-up replacement. What should the architect recommend?
- An airline classifies delay codes in a night batch and answers passenger questions during the day. The jobs need different models, and both may run inside the same data boundary. What should the architect use?
- An airline's prebuilt disruption copilot already rebooks routine delays. The only gap is a local crew-rest message the product can accept as added content. A team wants a full custom build of the same copilot. What should the architect choose?
- A consultancy's client already owns a copilot that can file status reports once it is configured with the client's template. The project team has opened a build estimate for a new agent. What should the architect do first?
- A hospital sends short coding suggestions and long nurse-note summaries through one oversized model. The suggestions cost as much as the summaries. What should the architect introduce?
- A retailer cleans product titles, which is simple and high volume, and also gives planogram advice, which is not simple. The team will not maintain two permanent single-model apps. What deployment fits?
- A university team hardcoded the largest model name into every campus bot, from a one-line room lookup to a multi-step aid appeal. What should replace that hardcoding for this mixed workload?
Design AI-powered business solutions (30-35%) · 105 questions
- A hospital network wants an agent on the patient portal that answers questions such as what to bring to a scan, using only the approved prep sheet, and that must not change the appointment or the chart. Which agent design fits?
- A university help agent uses classic topic matching. Its handbook is already connected as knowledge, but any question that misses a trigger phrase gets the fallback message that the agent does not understand, and the handbook is never consulted. What should the architect change?
- A distributor wants collection emails drafted from overdue invoices and a plain-language reason when a purchase order is late. Both answers must use the same customer and order records in Dynamics 365 Finance and Supply Chain Management. What should the architect orchestrate?
- A logistics company writes a shipment-exception row into Dataverse whenever a delivery fails. An agent should rebook a carrier from that row and should ask a dispatcher only when the new cost is above an approved limit. Which design fits?
- A hospital agent must send every appointment-cancellation request into one existing topic and must extract the appointment identifier. Patients phrase the request in many ways, and trigger phrases keep missing them. Free-form answers are not acceptable for this action. Which understanding approach fits?
- Clinicians in a health system say encounter when they mean a patient visit. Copilot in Dynamics 365 Customer Service answers as if they meant a sales meeting. The word is correct in the clinical glossary and wrong in the copilot's replies. What should the architect design?
- A utility must pull structured fields from scanned inspection forms and flag field photos that show an unsafe condition. Clerks currently retype the forms. Which capability should the architect propose?
- An insurer's agent must read a claim note and label its severity from written instructions. A separate, already designed flow then creates the reserve in the policy system. Which component should perform the severity labeling?
- A bank wants an agent that opens a card-replacement case and fills the required fields from the customer's request. It should not research unrelated products. Which design fits?
- A hotel wants an agent that posts a wake-up request to the property system when a guest asks for one, and that does nothing else. Which design fits?
- A professional-services firm wants an agent to file time entries in a billing system that has no API and no connector. Staff can enter time only through the system's browser screens. Which extension should the architect choose?
- An airline records a crew timeout when a duty runs long. An agent should start the legal rest recalculation from that record and stop for a crew scheduler if the next duty would break the rest rule. Which design fits?
- A bank wants an agent that explains what a fee means, using only the approved fee schedule, when a customer asks. The agent must not waive the fee or change the account. Which design fits?
- A museum wants an agent that answers visitor questions such as today's opening hours from the approved visitor sheet, and that must not book tickets or change a reservation. Which design fits?
- A pharmacy writes a refrigerator-alarm row when a drug fridge goes out of range. An agent should open a waste ticket from that row and text the pharmacist only if the drug is a controlled substance. Which design fits?
- A retailer wants an agent that issues a return label when an order meets the published return rule, and then stops. It should not recommend products or watch the warehouse overnight. Which design fits?
- A university wants an agent that submits a parking appeal the student has already completed, and that does no other work. Which design fits?
- A university wants an agent that tells applicants whether a program is listed in the catalog when they ask, and that must not offer admission. Which agent design fits?
- An airline baggage agent apologizes whenever a customer says bag or luggage. Those words are trigger phrases on the fallback topic, and the baggage guide is already connected as knowledge. What should the architect change?
- A retailer's fallback message tells a joke when the agent cannot answer, and brand reviewers rejected the joke. Questions that no topic and no knowledge source can answer still need a path. What should the architect change?
- A utility receives a meter-tamper event from the field network. An agent should open an investigation from that event and should not wait for someone to start a chat. Which design fits?
- Customers of a retailer ask where an order is in hundreds of phrasings. Each phrasing must open the same tracking topic and capture the order id. Which understanding approach fits?
- A bank kiosk agent only needs to recognize a small, stable set of exact commands such as balance and hours, then open the matching topic. Which understanding approach fits?
- A hospital agent's topics use one-word triggers, so a question that should miss still enters a topic. Questions that truly match nothing still need an honest miss path. What should the architect change?
- A retailer refunds an online order when the amount is under a published threshold and the item is unused. That path follows the rule and does not ask a model to judge the shopper. Which component should run the refund?
- A utility must send one approved planned-outage notice to a fixed set of accounts. An engineer, not the agent, decides whether that outage is safe to schedule. Which design fits the notice?
- An insurer already knows the claim severity. Posting the reserve is then a fixed sequence of connector calls, with no further judgment. Which design fits that posting step?
- Hotel guests ask open questions that the team cannot list as topics in advance, and a prose answer drawn from the property guide is an acceptable reply. Which orchestration approach fits?
- A bank wants a dispute summary written in the customer's own words. A flow already creates the case in the servicing system. Which split should the design use?
- A ward already has a flow that files the handoff note in the chart. Nurses want their shorthand turned into that structured note. Which component should do the rewriting?
- A clinic sends one approved appointment reminder, from a fixed template, to every patient on a fixed list the day before the visit. No sentence is generated. Which component should send those reminders?
- A retailer maps a shopper's free-text return reason to one code from a closed list. The list can sit in the instructions, and the shopper's text is the input. Which design should assign the code?
- A bank wants to flag emails that contain the word mortgage so a queue can review them. No other judgment is required. What should the design use?
- Catalog models in Microsoft Foundry already score most fraud types for an insurer, and they miss one rare label. The agent should call a model for that label. What should the design specify?
- A hospital will train a custom model in Microsoft Foundry to suggest discharge codes. Clinicians can label a curated note set. The live notes store changes all day and those notes are unlabeled. Which training design should the architect approve?
- A bank wants an agent to read account flags on a mainframe. That mainframe publishes no Model Context Protocol server and no API. A designer still asks to select the protocol this quarter. What should the architect decide?
- Standard vision labels miss one weld defect a manufacturer must catch. Photos from a second plant were kept out of training. What should the design call for?
- Consultants want a statement-of-work draft from the project files they personally can open, inside the Microsoft 365 Copilot they already use. The draft must stay limited to the person who asked. Which design fits?
- Faculty want a Microsoft 365 Copilot agent that summarizes the course folder a professor owns. Those professors can also open student mail about grades, and the summary must leave that mail unread. Which design fits?
- A retailer's inventory service already exposes a Model Context Protocol server with tools for on-hand quantity and bin location. A browser page shows the same numbers. The agent needs those quantities. Which extension fits?
- A consultancy is designing a Microsoft 365 Copilot agent for a law-firm matter. The agent may use files on that matter's SharePoint site and must not search the rest of the tenant. Which scope should the architect set?
- Two logistics agents both need shipment status, reroute, and proof-of-delivery actions. The platform team can publish those actions once on a Model Context Protocol server. Each agent team would rather paste its own copy of the API logic. What should the architect require?
- A hospital nurse line answers refill-status calls from people who are on a phone and cannot type. Which design fits the channel and the question?
- A city tax office must post property assessments in a legacy app that offers screens and no API. A person must review each posted assessment. Which design fits?
- A pharmacy must place wholesaler orders on a portal that is web-only this quarter. The vendor has promised a connector later. How should the design treat Computer Use?
- Hospital administrators want a Microsoft 365 Copilot agent that answers from the admin policy site. A teammate wants it grounded on the clinical archive so replies can quote chart detail. Which knowledge scope should the architect approve?
- An airline's Model Context Protocol server offers tools to read a booking, read a fare rule, and cancel a flight. The passenger-facing agent may explain a booking and must not cancel a flight. Which tool exposure should the architect approve?
- An airline hotline tells callers the new gate after a gate change. Callers are on the phone, and the only fact required is the current gate record. Which design fits?
- A distributor wants collection priority for customers with overdue invoices. Those invoices already live in Dynamics 365 Finance. Which design fits?
- Fraud analysts compare several transactions and write findings while the case file stays on screen. Which design fits this investigation?
- A retailer needs a vendor-hold explanation in the finance app and a purchase-order delay explanation in the supply chain app, and both must use the same company data. Which design fits?
- A utility outage line speaks with callers who report a dark house. The agent must confirm the address and log the outage. Which design fits?
- A hotel chain wants an agent to read room status from a property system that already has a supported connector. Which extension should the architect design?
- A business-to-business retailer wants the service case to open with the related order. Sellers do not work those cases. Which configuration fits?
- A consultancy wants seller meeting prep to use the opportunity. A teammate also wants a service copilot deployed to those same sellers. Which configuration fits?
- A hospital wants the customer-service copilot to draft the case summary and wants the contact-center copilot to use that case. A separate FAQ bot has been proposed. Which design fits?
- An insurer's sales copilot and service copilot both answer coverage questions, and they sometimes cite different policy numbers for the same customer. Which design fits?
- A manufacturer's buyer wants a summary of a purchase order whose lines are still waiting on receipt. The order is already open in Dynamics 365 Supply Chain Management. Which design fits?
- A university's recruitment copilot and student-service copilot have quoted two different application deadlines for the same program. Which design fits?
- On a bank service desk, staff say instrument and mean a loan product. The customer-service copilot answers as if they meant a musical instrument or a generic tool. What should the architect add?
- A hospital wants the customer-service copilot's case summary to lead with allergies before the rest of the case. A teammate offers a second summarizer that staff would open instead. What should the architect do?
- An airline's service copilot offers a recovery before it knows the member tier stored in the customer app. Which design fits?
- A retailer wants the service copilot's reply to include the customer's loyalty id. A proposal builds a new agent and hides the in-app copilot so staff must use the new one. Which design fits?
- An insurer seats claims adjusters and insurance brokers in one building. Adjusters work cases all day, and brokers work pipeline and meetings. A sponsor wants one copilot product for both desks. Which design fits the two roles?
- A manufacturer turns on Copilot for Sales for its dealers, but the dealer price list is not connected, so quotes ignore contracted prices. What is the right configuration?
- Passengers call an airline's Dynamics 365 Contact Center voice line about missed connections, and almost none of them open the website chat. Where should the agent be designed to meet them?
- Sellers at a manufacturer ask Copilot in Dynamics 365 Sales for a dealer price. That price list lives in a system outside Dataverse, and the copilot cannot see it. What should the architect add?
- A university registrar forbids a casual tone, but the customer-service copilot's suggested reply still sounds casual. A teammate wants that copilot switched off. What should the architect change?
- Airline agents type irregular operations into the customer-service copilot and receive an answer about an ordinary delay. At this airline the phrase means a disrupted operation with its own recovery steps. What should the architect define?
- An insurer wants Copilot in Dynamics 365 Sales to help a broker prepare for the next appointment. A proposal connects a copy of the entire claims database to that copilot. What should the connector bring?
- A bank's Dynamics 365 Contact Center agent verifies the caller and must then pass the call to a person who should not have to hunt for the case. What should that handoff include?
- Sellers at a consultancy ask Copilot in Dynamics 365 Sales whether a deal still clears the margin rule. The figures live in the project-accounting system, and a teammate pastes them into the opportunity notes. What should the architect connect?
- A utility's customers report outages by SMS in Dynamics 365 Contact Center, but the proposed agent is designed only for the email queue. Which design fits those customers?
- A retailer wants store exceptions on a custom page in the model-driven app, and a manager must accept the agent's proposed stock transfer inside that same app. Which design fits?
- A retailer must write alt text for product photos and must block photos that contain unsafe imagery. Which design fits?
- Underwriters at a bank must review agent tasks that are waiting, and they do that work inside the loan's model-driven app. What should the design give them?
- An insurer's claim calls arrive as audio, and adjusters retype each call before an existing text agent can read it. What should the design propose?
- A bank splits monthly statements into chunks before they ground a service agent. A retrieved passage sometimes shows one customer's lines under another customer's name. What should each chunk keep?
- Hospital clerks need a custom bed-board page inside the model-driven app they already use, built as code against Dataverse, without rebuilding the rest of that app. Which design fits?
- Utility field technicians work in a canvas app. An agent can suggest a replacement part, but the technician must accept it, and the app must remain the system they use. Which design fits?
- A hospital wants clinical notes to ground a chart-summary agent, but some scans still show another patient's sticker in the margin. What should the design do with those notes before grounding?
- A utility wants inspection PDFs to ground an agent. The files are raw scan dumps that mix inspection pages, stamps, and blank sheets. What should happen before grounding?
- A pharmacy canvas app is the counseling checklist a pharmacist must finish. The team also wants an agent summary of the visit. Where should that summary be requested?
- A retailer is building a canvas app for store inspections. An agent should comment on the visit only after the required photos are attached. When should the app call the agent?
- An insurer is connecting a new claim system. The existing Copilot Studio agent already holds the topics adjusters use. A proposal would create a second agent and copy those topics into it so the new system has a home. What should the architect do?
- Shoppers abandon a retailer's product agent because every reply is a wall of text. The team wants to stop that drop-off. Which change matches the gap?
- A university student agent answers registration questions. Staff are pasting the parking rules into every topic so each topic can recite them. The parking system already exposes a tool that looks up a permit. What should extensibility mean here?
- A hospital intake agent stops when the model endpoint is unavailable, and the clerk has no other path to continue the visit. Which gap is this, and what should the design add?
- A bank's agent writes the full account number into the run log so analysts can read it later. Which Well-Architected pillar is missed, and what should change?
- Maintenance crews at a manufacturer already coordinate in a Teams channel. The agent there starts a new email thread whenever a fault is reported. Supervisors want the work order opened in the channel the crew is using. What should the agent do?
- Consultants on an engagement already work in one Teams channel, and the project files live on that project's SharePoint site. A proposal would host the project agent on a public website with a separate library. Where should the agent meet the team, and what should ground it?
- An airline's Copilot Studio agent already books seats and answers fare questions. Catering wants that same agent to book a special meal, and a teammate proposes a separate copy of the agent for each meal type. What should the architect do?
- A hospital wants a SharePoint agent that answers nursing policy questions. The draft connection lets the agent read every site in the tenant, including finance and executive sites the nurses do not use. What should the architect change?
- Agency caseworkers want answers while they read the case document in SharePoint. A proposal would send them to a separate site that has its own login. What should the architect propose?
- Retail buyers spend the day in email and want a daily brief of vendor messages. A proposal would build a new website and ask buyers to upload those messages before they can read a brief. What should the architect propose first?
- Buyers need the finance and operations agent chat to apply the vendor-compliance manual. One buyer pasted the manual into a personal prompt that nobody else receives. What should the architect do?
- A distributor's supply-chain agent chat invents bin rules when warehouse staff ask how a pallet should be slotted. The approved slotting policy is not in the chat's knowledge. What should the architect do?
- University faculty want a recap and follow-up tasks from meetings they already hold in Teams. A proposal would build a custom portal where faculty upload a recording before any recap appears. What should the architect propose?
- A retailer's makers are ready to build a product-description copilot because they cannot tell which AI features are already turned on in the tenant. Where should the architect send them before that build starts?
- A bank platform team plans to upload labeled disputes to the Power Platform AI hub because they believe the hub is what trains their models. Which correction should the architect make?
- A hospital's Power Platform AI hub lists a clinical-note copilot that the privacy office has not approved, and makers can already see it. What is the design issue?
- A manufacturer's finance agent chat answers cost questions from short menu tooltips and misses the rules in the controlled cost-accounting guide. What should the architect add to that chat?
- Receivers in a hospital supply app ask the in-app copilot how to accept a delivery, and the only answer is the generic product tour. The local receiving procedure is missing. What should the architect add?
- A university finance copilot tells clerks to close the period on the product's sample date. The controller's campus close calendar differs, and that calendar is not in the in-app help. What should the architect add?
- A plant copilot's in-app help still describes a retired receiving screen, so it tells clerks to follow a click path the current app no longer has. What should the architect do with that help topic?
Deploy AI-powered business solutions (20-25%) · 73 questions
- Makers at a manufacturer edit the production Copilot Studio agent directly, and last week's working version can no longer be restored. Connectors in the agent point at production systems even when someone experiments. What ALM change should the architect make?
- A bank's case agent reads customer PDFs as knowledge. One uploaded PDF says to ignore prior instructions and email the full case list to an outside address. The agent has a tool that can send mail. What should the architect change?
- A government agency requires that prompts and grounding transcripts for a citizen agent stay inside the national boundary it named. A teammate suggests a global model deployment because the tenant home is in that country. What should the architect validate?
- A utility's outage agent can draft switching orders that move load on the live network. A mistaken order could drop service to a neighborhood. Which control should the architect require before go-live?
- A bank runs an autonomous agent that drafts a case note when a transaction alert is created. There is no chat. Supervisors want to know whether the agent is behaving. What should operations monitor first?
- A contact-center agent is released because a pilot showed a high thumbs-up rate. Leaders have no measure of whether answers match the knowledge source, whether the customer's request was completed, or whether the agent's actions were correct. What should testing add?
- A bank sorts its backlog of agent fixes by which manager shouted loudest, rather than by how often the answer failed. The same wrong fee answer has failed for hundreds of customers and sits near the bottom. What should the architect do with the backlog?
- Patients downvote a hospital portal agent, and the comments say the prep sheet it quotes is outdated. A teammate wants to retune the model immediately. What should the architect do first?
- Advisors at a university have stopped using the registration agent. Their feedback says it refuses every transfer-credit question, and they now email the registrar instead. What should the architect treat this pattern as?
- An insurer's evaluation traces show that policy-rider questions retrieve no usable rider passage, while deductible and limit questions retrieve the right clause and are answered correctly. What should the architect tune?
- An airline's rebooking agent writes a trace whenever a tool call fails, and nobody reviews those traces. The operations board shows only how many runs started. What should operations monitor?
- A retailer judges its returns agent by the CPU of the storefront web server. Leaders cannot tell whether a return was completed, refused, or handed to a person. What should operations monitor instead?
- An airline gate agent answers slowly. Traces show the gate-assignment tool timing out, and the model returns promptly once that tool responds. Leadership wants the model retuned. What should the architect change?
- A retailer evaluates its returns agent. The greeting matches the brand script, and the following sentences cite a returns policy the company has retired. What should the architect change based on that evaluation?
- A manufacturer's agent is slow at only one plant, and only on the turns that call the parts service. Turns that do not call that service return quickly. What should the architect inspect before any model change?
- Leaders at a utility want one number for the outage agent, which tells a customer the current status of that customer's outage. Which metric should the architect report?
- A hotel booking agent shows a high action count. Traces show the booking tool failing and the agent retrying the same call. Which metric should operations use to judge this agent?
- A bank reports the average length of its fee agent's replies as the quality number. Supervisors need to know whether the fee in the reply matches the fee schedule. Which metric should replace the length figure?
- A logistics agent drafts delay notices. Traces show the model returns quickly, while the carrier-status tool accounts for most of the wait on each run. What should the architect change first?
- A hospital portal agent starts refusing more questions in the hour after a handbook is published into its knowledge. Telemetry records both the publish and the refusals. What should operations examine first?
- An insurer wants to promote a custom fraud model. Nobody has agreed how precise it must be, and no holdout set exists. What must be in place before production?
- A public agency's eligibility agent must include one required statutory sentence, word for word, in every eligibility answer. Current tests pass when the reply is polite. What should the suite add?
- A pharmacy agent's tests pass whenever any reply comes back to a symptom question. Pharmacists require that the agent not recommend a dose. What should the test suite add?
- A hospital validated a clinical-coding model only on notes from one clinic and wants to use it at every site. What should the validation criteria require before that wider use?
- At a distributor, the sales copilot drafts a quote, the finance copilot creates the invoice, and the service copilot opens a case. Each copilot looks right in its own app, yet the case names a different total than the invoice. What should the architect require?
- A manufacturer validated its weld-defect model only on sharp studio photos. Plants submit glare, dust, and cropped beads, and none of those photos are in the validation set. What should the architect require before promotion?
- A bank's approved prompt rules allow a copilot to summarize a credit file and forbid any prompt that directs staff to decide a lending exception. A submitted prompt says to decide the exception and reply approved or denied. What should prompt review do?
- A retailer submitted a copilot prompt that pastes the entire returns manual into the instruction text. Staff already keep that manual as a governed document. What should prompt review decide?
- A university's approved prompt rules require a refusal line when a question is outside the student handbook. A submitted study-skills prompt sets a role and an output shape, and it has no refusal line. What should prompt review do?
- A hospital books a supply request in one app, and finance places that same request on a hold list when the amount needs review. Tests of each screen pass on their own. What test should the architect add before go-live?
- A bank overwrote the production knowledge file for a product agent and has no prior copy to restore. What should lifecycle management require for that grounding data?
- A university edits the handbook file that grounds a help agent directly in production. How should lifecycle management treat that handbook?
- A retailer wants faster test coverage for a returns agent whose approved policy is already written. How should the suite be built?
- An airline loyalty desk changes a passenger's tier in its Dynamics 365 sales app and must see that tier on the open service case. Which test should the architect require?
- A hospital tests a prep-sheet agent that may answer only from the approved sheet. How should the test cases be produced?
- A bank uses Copilot to draft test cases for a payment agent. The draft covers successful payments and skips declines and timeouts. What should the test strategy do?
- A utility is ready to release a Microsoft Foundry agent to production, and the release path has no evaluation gate. What should the architect add?
- An insurer's test agent and production agent both read one policy library, so a wording edit made in test shows up in production answers the same day. What should the architect change?
- A hospital maker edits the production Copilot Studio agent and exports that unmanaged change to save it. What should the architect require instead?
- A bank team edits a Microsoft Foundry agent directly in the production project. What should the architect require instead?
- A retailer imports a Copilot Studio solution into test, and the connection references still point at production. What should happen before the next promote?
- An airline promotes its Microsoft Foundry agent and the model deployment on separate schedules, and the two have drifted. What should the release do?
- A hospital's test environment and its production environment call the same model deployment. A test prompt changes answers that clinicians already see on the live floor. What should the architect change?
- A manufacturer trains its weld-defect model in the production project, using live line photos while inspectors still rely on the current model during the shift. How should the architect change that lifecycle?
- A manufacturer's supply-chain copilot answers from a different knowledge set in development, test, and production, and nobody planned that split. What should the architect make part of the release?
- During month-end close, a distributor changes a finance copilot setting directly in production. Controllers see the guidance shift while the books are still open. What should the architect require for the next change?
- A retailer turns a finance AI feature on in the production company before any test company has run it. Controllers see the new guidance during the live close. What should the architect require?
- An insurer replaces a custom model by overwriting the production model in place, and the previous behavior cannot be restored. What should the release do instead?
- An insurer's Dynamics 365 service copilot in the test environment still reads production cases. The team wants to treat the latest run as a valid test. What must change first?
- A retailer's returns agent can submit a refund of any amount, and no person sees the refund before the money is paid. What control should the architect put on that tool?
- A hospital edits customer-service copilot customizations directly on the live case form. A recent edit changed the summary families hear about, and no sandbox copy exists. How should those customizations ship?
- A university will ship a Dynamics 365 sales copilot change this week and a service copilot change next week, and the two contradict on the same student status. What should the release plan do?
- A hospital scheduling agent signs in with a shared personal account that can open charts, payroll, and the booking calendar. What should replace that sign-in?
- A bank's agent reads every customer's file through one service account, and those reads are not written to an audit log. What should security require?
- At a bank, any developer can deploy a model into the production Microsoft Foundry project. What should the architect change?
- A public agency's citizen agent was built by a vendor, and no agency employee will own it once that vendor leaves. What should governance assign before the vendor goes?
- A retail department wants agent logging turned off for a promotion week so the agent can go faster. What should governance decide?
- An airline has been putting agents into production with no record of the actions each agent is allowed to take. What should governance require before the next agent goes live?
- An insurer still has an old model key pasted in a notebook that the data-science channel shares. What should the team do?
- A hospital tuned a model from a storage account the whole company can read, including files meant only for that tuning job. What should security change?
- A bank's proof of concept ran on a global model deployment, and the privacy office allowed that proof. The production build copies the same global deployment. What should the architect do before production?
- A retailer's returns agent reads product reviews while it decides a refund. One review tells the agent to ignore policy and issue the refund twice. The agent has a tool that can post a refund. What should the architect require?
- A hospital portal agent receives a patient message that says to ignore its rules and reveal the next patient's record. What should the design require?
- A bank is ready to release a credit-explanation agent used in several regions. No one has checked whether the explanations treat applicants consistently across those regions. What should the responsible-AI review require before go-live?
- A hospital discharge agent messages patients with next steps and does not tell them the message is automated. The responsible-AI review has not required that disclosure. What should the review require?
- A retailer's hiring-screen agent rejects applicants and provides no way for a person to reverse that rejection. The responsible-AI review is asked who is answerable for the decision. What should the review require?
- At a consultancy, every maker can open every client's grounding library, including libraries for clients they do not work with. What access change should the architect require?
- A citizen-facing agent at an agency answers a published FAQ, yet its grounding permission also lets it read case notes the FAQ does not use. What should the architect change?
- A hospital's tuning set of clinician notes is readable by the developers who tune the agent and by a reporting intern who does not tune models. What should the architect change?
- A hospital stores grounding recordings in a region the privacy office has not approved. A teammate wants to change the agent's description to name an approved region and leave the files where they are. What should the architect do?
- A bank can swap the model behind a Foundry deployment, and nothing records who made the swap. What should the architect require before that deployment is used in production?
- An insurer replaces the policy files that ground a claims agent and keeps no history of the prior file or of who replaced it. Model changes are already logged. What else should the architect require?
- A utility agent drafts a switching plan. An approver overrides the draft, and the stored record keeps the original draft with no sign of the override. What must the audit trail include?
These questions are original practice material and are NOT actual exam questions or brain-dump content. Microsoft and related marks are trademarks of their respective owners. This material is not affiliated with, endorsed by, or sponsored by Microsoft.