Microsoft Certified: Azure AI Apps and Agents Developer Associate practice questions
Microsoft · AI-103 · 200 questions
Original practice questions for Microsoft Certified: Azure AI Apps and Agents Developer Associate.
This course contains the use of artificial intelligence.
About the AI-103 exam
- Time allowed
- 2 hours
- Languages
- English, Chinese (Simplified), Chinese (Traditional), French, German, Japanese, Korean, Italian, Portuguese (Brazil), Spanish
- Format
- Proctored, may include interactive components; Pearson VUE
Schedule this exam The certification this earns
Exam details published by the vendor, checked 25 August 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
Browse by Domain
Study specific topics at your own pace.
Plan and manage an Azure AI solution (25–30%) · 55 questions
- A city permits office needs a simple FAQ bot that answers text questions about zoning rules and never processes photos or scanned plans. Which Foundry model choice best fits that generative task?
- A county 311 team wants resident answers grounded in published open-data PDFs rather than freeform model guesses. Which planning decision best meets that requirement?
- A transit authority needs an agent that reads service-alert knowledge and calls live bus ETA APIs for riders. Which Foundry service mix should the AI developer plan?
- A public-health department must analyze clinic photos together with clinician text notes in one workflow. Which Foundry choice best matches that multimodal civic workload?
- A housing authority indexes lease PDFs so a resident assistant can answer questions from those documents. Which retrieval and indexing approach should the team select?
- Emergency management wants an agent that remembers prior incident briefings across shift handoffs. Which Foundry planning choice addresses that need?
- A library system needs a staff agent that can call the catalog API and also answer from a policies knowledge base. How should tool and knowledge services be matched?
- A parks department only needs seasonal brochure copy generated from outlines—no tool calling or multi-step agents. Which Foundry path should they choose?
- A clerk of courts wants a resident bot that finds answers in docket guides by meaning, not only exact keyword matches. Which grounding search style should they plan?
- A water utility wants form OCR enrichment before an agent answers billing questions from scanned statements. Where do Foundry Tools fit in that plan?
- School-district public IT needs an on-budget classroom-help bot with tight latency targets. Which model class should they favor for that constraint?
- A city open-data portal must index both CSV dataset metadata and PDF methodology notes for a resident search assistant. Which indexing plan fits that heterogeneous content?
- A fire department agent must consult a radio-log knowledge store and call a CAD status API. Which Foundry design best combines those needs?
- The mayor’s office wants multimodal models for press-photo captioning but text models for legislative Q&A. Which planning principle should guide that split?
- County IT must design Azure infrastructure for a Foundry project that hosts a permitting chatbot and shared models. What should that design emphasize?
- City DevOps needs a deployment option for a seasonal parks chatbot that spikes on holiday weekends. Which approach best fits variable civic load?
- A transit team is configuring a rider-assistant inside a Foundry project. Which setup step is essential?
- A housing authority wants Foundry project builds to ride the existing municipal CI/CD pipeline. What integration should they implement?
- Public works wants clear separation between a sandbox Foundry project and the production resident-facing agent. Which environment design is appropriate?
- A clerk’s office will share one summarization model across multiple department apps. Which Foundry deployment pattern fits?
- Emergency ops needs low-latency agent responses near the Azure region that hosts CAD integrations. Which infrastructure consideration matters most?
- A library configures an agent deployment so it can reach only approved catalog and hours APIs. What configuration practice does that reflect?
- A city platform team must document how the 311 web app connects to its Foundry project. Which setup concern is central?
- A health department will promote a tested agent from staging to production using pipeline gates. Which CI/CD practice applies?
- Parks plans deployment capacity before launching a summer festival chatbot expected to draw heavy traffic. What should capacity-aware deployment include?
- A county security review for a Foundry-based permitting assistant requires AI traffic to stay off the public internet between Azure resources. Which design choice should the platform team include?
- A water utility runs Foundry agents for both field operators and a public billing chatbot. What setup best separates those audiences?
- An open-data team’s Foundry generative app depends on a civic search index whose schema changes with each data release. How should they keep the app aligned?
- After a major storm, a city’s resident AI assistant spikes in usage and FinOps sees token spend and throttling. What should the AI platform team manage first?
- A transit authority updates bus-route PDFs, then riders report wrong answers from the Foundry RAG assistant. What should operations monitor?
- A public-health FAQ agent starts missing newer vaccine-policy wording that residents now use. What monitoring concern does this raise?
- 311 operations notice abusive and unsafe messages hitting the city’s public chatbot. Which operational practice should they emphasize?
- A housing authority bulk-uploads lease PDFs and the job fails halfway. What should the AI ops team monitor for the resident assistant’s retrieval layer?
- A clerk of courts agent must call Azure resources from Foundry without embedding long-lived secrets in code. What credential approach should they configure?
- A county runs a confidential casework bot on Foundry and must keep the endpoint off the public internet. Which control should they apply?
- A library system wants volunteers to help draft content but not redeploy production Foundry agents. What should administrators apply?
- Emergency management runs tabletop drills that hammer the same Foundry models used by the live public alert bot. How should they protect the live workload?
- After a tariff change, a water utility’s billing FAQ retrieval starts returning outdated policy chunks. What should AI ops monitor?
- A city SOC sees a sudden spike in tool calls from a Foundry agent tied to a service principal that may be compromised. What capability does this illustrate?
- Parks expects a weekend concert series to spike chatbot traffic and wants capacity without a permanent year-round cost jump. What approach fits?
- An open-data portal’s RAG pipeline begins producing empty embeddings and failed skillset runs. What should operators alert on?
- Courts IT still uses shared API keys across Foundry connectors and wants a safer pattern. What migration should they prioritize?
- A city is preparing a public 311 generative assistant on Foundry. Which responsible AI control should be configured before launch?
- A county must release a benefits-eligibility agent but leadership requires evidence it behaves safely. What should the team enable before release?
- A housing authority requires every denial-explanation answer from its Foundry assistant to be reconstructible later for appeal review. What should they require?
- A clerk’s filings agent can draft submissions but must not file documents without a human. Which responsible AI pattern fits?
- A transit rider assistant must call service-alert and ETA tools but must never reach internal payroll APIs. What control enforces this?
- A public-health mental-health resource chatbot must detect self-harm language and respond with safe redirection. Which responsible AI capability is required?
- The mayor’s office wants reviewers to see why the policy bot cited a particular ordinance section. What should the solution emphasize?
- A library patron assistant must refuse medical diagnosis requests and stay within library-service scope. Which control should be configured?
- Emergency operations wants an agent to draft public alert text, but a public information officer must approve it before release. Which responsible-AI control should be configured?
- A water utility caught a wrong shutoff-notice draft during review and needs to reconstruct how the agent produced it. Which responsible-AI practice should the team use?
- A parks events agent must not invent sponsorship deals that the city never approved. Which responsible-AI approach should be applied?
- An open-data team plans to enable image upload in a multimodal civic app. What should they run before expanding that feature?
- Courts IT must keep a public FAQ agent from using the same mutable docket tools available to an internal staff agent. Which governance approach fits?
Implement generative AI and agentic solutions (30–35%) · 65 questions
- A city wants residents to use a permitting pre-check chat before filing. Which Foundry generative step should developers implement first for that experience?
- County 311 needs high-volume address-normalization suggestions with tight cost and latency budgets. Which generative model choice fits best in Foundry?
- Transit staff need help turning GTFS snippets into reusable service-alert templates. Which Foundry model type best supports that assisted authoring?
- Parks wants a generative app that drafts accessible descriptions of trail-map images for the website. Which Foundry consumption pattern fits?
- A housing authority needs the resident bot to cite current occupancy rules from policy PDFs rather than invent answers. Which generative pattern should they implement?
- A clerk's office needs a generative flow that classifies a filing type, drafts a checklist, then cites applicable fees. Which application design fits?
- Public health wants a generative reply that first pulls live clinic hours from an API, then drafts the resident message. Which pattern should developers build?
- A library catalog bot sometimes invents book holds that do not exist. What should the team evaluate for in the generative app?
- After refreshing the storm-debris FAQ index, 311 wants to know whether answers still match resident questions well. What should they measure?
- Emergency management plans to launch a public preparedness chatbot. Which evaluation should they complete before go-live?
- A water utility billing portal must call the generative service from its existing web application. Which integration approach fits Foundry?
- An open-data portal web app must reach the correct Foundry project that hosts the generative assistant. What should administrators configure?
- Courts staff need ordinance summaries streamed from a model that is already deployed in Foundry. What should the staff application do?
- A city grants portal must answer eligibility questions from published rulebooks, not from unverified model guesses. Which pattern should the team implement?
- Transit wants a generative workflow that drafts delay notices and then waits for PIO tool confirmation before publishing. Which design fits?
- Housing staff worry the RAG bot cites lease clause IDs that do not match retrieved documents. What should they evaluate?
- Parks wants a seasonal copy generator that sends drafts into the department CMS through an API. Which Foundry generative-app approach fits?
- County 311 is switching the bot to a newer multimodal deployment and needs assurance before full cutover. What should the team do?
- A clerk needs a generative step that emits structured JSON fields the fee calculator can ingest. Which generative-app technique should be used?
- Public works wants a field-supervisor coaching chat grounded in standard operating procedures rather than improvised advice. Which pattern should they implement?
- A city library wants a Python service that answers storytime-program questions using a Microsoft Foundry project. Which integration pattern should the developer implement?
- A mayor’s office must evaluate both safety and tone quality of a public town-hall summary generator before wider release. Which approach best fits that combined evaluation?
- A city is building a permitting agent that must stay in a filings-help role, pursue a clear goal, and track multi-turn filing conversations. What should the team define first in Foundry agent design?
- A county 311 team wants an agent that can create service requests through an API. Which design step is essential for reliable function-calling?
- A transit authority needs an agent that answers from service bulletins and also calls a live ETA function. Which pattern should the developer implement?
- A housing authority agent must remember a resident’s open work-order context across multiple turns. Which capability should be implemented?
- A city clerk needs an agent that uses filing instructions from a knowledge store and looks up current fees from an API. How should those capabilities be exposed?
- A library agent must read a scanned community-event flyer before answering patron questions about the event. Which tool pattern is appropriate?
- Emergency operations wants intake, logistics, and public-information draft work handled by cooperating specialists rather than one overloaded bot. Which design should they implement?
- A water utility wants an agent that drafts shutoff notices but must not send them without a supervisor. Which agent pattern should they use?
- Parks staff see a deployed events agent giving wrong venue answers and need to improve it. What should they do next?
- An open-data portal agent must query the city’s dataset metadata catalog through a controlled interface. Which tool approach is most appropriate?
- A courts public FAQ agent must help residents find information but must never mutate case records. How should its tools be constrained?
- A public-health agent walks residents through a triage flow and then must connect them to a nurse line. Which combination should the design include?
- A 311 system uses a classifier agent and a work-order agent that must share intake context. Which orchestration pattern fits?
- A transit agency is piloting an autonomous agent that suggests service reroutes but must not post to public channels unchecked. What should the workflow include?
- A housing authority wants its agent to cite eligibility policy instead of improvising rulings. How should agent goals be written?
- A clerk agent’s fees API sometimes times out. What should the team design after analyzing those tool failures?
- Library staff need an agent that searches an internal wiki and can book meeting rooms through an API. Which tool mix is correct?
- A fire department wants a runbook assistant where research work stays separate from dispatch suggestions. Which multi-agent choice supports that?
- City IT just deployed a permitting agent and must observe traces, failures, and usage in production. What should they add?
- A county benefits agent must check whether an ID number is valid against an eligibility API before continuing. Which technique should it use?
- A parks planning agent must check weather and the facilities calendar before advising on outdoor event feasibility. What should its tool design include?
- The mayor’s office wants an agent to draft replies to batches of constituent email but not send them unsupervised. Which control is required?
- A city wants its permitting bot to stay concise and avoid sounding like legal counsel. Which optimization should the team apply?
- A county benefits bot drafts eligibility summaries. Which optimization should the team add before showing a final summary to a resident?
- A transit authority needs to see where rider-agent delays and token spend come from after each release. Which capability should the team implement?
- A housing authority wants cheap FAQ answers for routine questions but a stronger model for complex lease interpretations. Which optimization pattern should they use?
- A clerk of courts needs fee quotes that never invent amounts outside the statutory fee table. Which orchestration approach should the developer implement?
- After a prompt change, a 311 team wants to know if unsafe replies rose even when latency looks fine. What should their observability include?
- A public-health department wants richer critique of triage drafts without exposing long reasoning chains to residents. Which approach fits?
- A city library's program assistant is overspending tokens on long prompts and oversized retrieval chunks. Which optimization should the team apply?
- A water utility shares one Foundry deployment across departments and needs fair cost allocation. Which operational capability should they enable?
- Emergency management wants alert severity decided deterministically, then rewritten in plain language for the public. Which pattern should they implement?
- Parks staff keep getting overlong brochure drafts from their generative app. Which model-parameter change best addresses that?
- An open-data portal's RAG answers feel slow. What observability step helps the team find whether retrieval or the model is the bottleneck?
- Courts IT wants docket summaries shown to clerks only when citation IDs are present. Which generative optimization helps catch missing citations?
- A city portal routes simple how-to questions cheaply and complex policy questions to a stronger model. Which optimization describes this?
- A housing work-order agent sometimes fails when calling maintenance APIs. What should operations dashboards track?
- Transit wants grounded service-change answers to always include the effective date. Which optimization should the team apply?
- A county finds that reflection loops improve 311 answer quality but push latency past the published SLA. What should the team do?
- A library wants seasonal program assistants to keep an approved voice when volunteers edit prompts. Which operational practice should they adopt?
- Public works needs outage maps from GIS to be authoritative, with the LLM only explaining the affected area in plain language. Which pattern fits?
- After enabling a new constituent-email agent, the mayor's office needs to watch for rising unsafe outputs. Which practice should they follow?
- A clerk's office wants one assistant experience that runs intake, then fee quote, then checklist generation as coordinated steps. Which optimization should they implement?
Implement computer vision solutions (10–15%) · 27 questions
- A parks department needs promotional images for a summer concert series flyer from a written description. Which computer-vision solution should they implement?
- Transit wants a short explainer video built from a script and approved station photos. Which generation approach should the team use?
- City communications must keep generated social graphics aligned with official logos. Which generation practice should they apply?
- Housing needs to remove a temporary construction fence from a generated community-room photo before publishing. Which editing technique should they use?
- A clerk must anonymize faces in a public meeting still before posting it online. Which image-editing approach should they use?
- Emergency management must replace an outdated shelter name card in a generated preparedness video before release. Which approach should they use?
- A library needs consistent kids-program thumbnails and must set style and aspect ratio before generation. Which capability should they use?
- An open-data team wants AI-generated visuals that illustrate a dataset story without depicting identifiable people. Which generation use is most appropriate?
- A water utility needs to highlight a main-break area on a generated map graphic for public notice. Which editing approach should they apply?
- A 311 system receives resident pothole photos and must understand the visual scene before routing the ticket. Which multimodal capability should they use?
- Parks must add concise captions to many trail photos in an accessibility gallery. Which multimodal workflow should they configure?
- Transit staff need to ask which platform sign appears in a station photo and get an answer grounded in that image. Which capability should they enable?
- A library must publish exhibit images with alt-text and extended descriptions that support accessibility guidelines. Which multimodal output should they generate?
- Courts need to extract visual characteristics from exhibit scans using Foundry tooling. Which approach should they configure?
- Emergency ops wants to interpret timestamps and actions in training-drill video segments. Which multimodal workflow should they implement?
- Housing wants a focused pipeline that checks unit-condition photos against a simple checklist. Which Content Understanding approach fits best?
- A clerk must process complex multi-page scanned packets that mix text and varied visuals. Which Content Understanding mode should they set up?
- Public works wants work-order enrichment that names objects such as hydrants, curbs, and cones in field photos. Which multimodal capability should they use?
- City communications needs richer captions than one-line blurbs for a batch of ribbon-cutting event photos. Which captioning choice should they make?
- Transit must interpret bus-interior camera segments for left-item reports under an approved policy. Which multimodal workflow should they apply?
- An open-data portal publishes chart images inside PDF annexes and wants staff to ask questions answered from those charts. Which capability should they build?
- A city must stop unsafe or disallowed images from entering the 311 photo-upload path. Which responsible-AI control should they implement?
- A county multimodal clerk bot may receive images that hide malicious instructions in embedded text. Which risk should the team detect and mitigate?
- Parks must mark AI-generated promotional images before they go public. Which visual policy control should they apply?
- Transit generates rider graphics and must catch prohibited symbols or unapproved branding before publish. Which policy enforcement should they add?
- A library gallery publishes makerspace photos submitted by patrons and must screen them first. Which multimodal safety step should they take?
- Housing uses a multimodal agent that can act on unit photos, but high visual-policy risk scores must stop those actions. Which control should they enforce?
Implement text analysis solutions (10–15%) · 26 questions
- A clerk needs party names and case numbers pulled from free-text hearing requests. Which text-analysis approach should they use?
- 311 receives long resident complaint emails and must turn them into structured topic labels for routing. Which text-analysis capability should they apply?
- Housing receives free-text maintenance reports and needs machine-readable fields for unit, issue type, and urgency. Which output pattern should they use?
- A public-health department wants to understand whether community feedback about new clinic hours sounds frustrated, supportive, or mixed before staff reply. Which text-analysis capability should they apply first?
- A transit authority needs rider feedback that mentions threats or safety risks escalated to humans before any automated reply. Which text-analysis step should the pipeline include?
- A benefits office must find Social Security numbers and similar sensitive patterns in chatbot transcripts before logs are retained for analytics. What should the text-analysis workflow detect?
- A city library system needs bilingual program descriptions published from English source text using a dedicated translation service in Foundry Tools. Which approach fits best?
- A mayor's office receives constituent emails that need nuanced, context-aware rephrasing into another language beyond straightforward phrase substitution. When should they prefer an LLM-powered translation flow over basic Translator usage?
- A clerk of courts must produce compliance-oriented summaries of lengthy administrative orders that always retain mandatory legal citations and omit casual commentary. How should they customize language-model text analysis?
- A water utility wants work-note extraction that understands meter-related jargon such as register freeze and AMI hop failures. What customization should they apply?
- An open-data team receives thousands of dataset suggestion emails and needs recurring themes identified for the product backlog. Which text-analysis capability should they run at scale?
- Emergency management needs a short EOC board brief built from many multi-agency status emails without requiring directors to read every thread. Which text-analysis action best fits?
- A parks department notices reservation cancellation messages growing sharper and wants an early signal of service problems. Which analysis should they apply to those messages?
- County counsel wants ordinance amendment drafts turned into structured JSON fields such as section id, change type, and effective date to support diff-review helpers. What should text analysis produce?
- A 311 call center wants residents to dictate service requests over a phone IVR that feeds an agentic intake flow. Which speech capability should they implement first?
- A transit rider app agent must speak gate-change announcements aloud inside the app when travelers cannot easily read the screen. Which speech capability should the team add?
- A public library wants its internet-PC assistant to accept spoken questions and reply with spoken answers for patrons who struggle with keyboard-only interaction. What should they enable?
- A clerk of courts finds that generic speech recognition mishears local docket jargon during dictation into the case assistant. Which speech approach should they explore?
- Emergency operations wants an incident assistant that reasons over radio-clip audio—not only transcripts—to surface situational cues for responders. Which capability should they enable?
- A public-health clinic must turn spoken discharge instructions into another language for patients during the same visit. Which speech solution should they implement?
- Housing maintenance staff need to dictate inspection notes that flow directly into the case agent for follow-up tasks. How should speech connect to the agent workflow?
- A parks kiosk agent must read trail alerts aloud for visitors who cannot rely on on-screen text alone. Which speech pattern should they implement?
- A city council AV team needs live multilingual accessibility drafts from spoken town-hall remarks for residents who use other languages. Which speech capability should they use?
- Field technicians at a water utility ask spoken questions about SOPs and need answers grounded in the approved procedures library. Which design should the agent use?
- An open-data team records podcast clips and wants newsletter drafts that start from speech, then apply summarization. Which pipeline order is correct?
- A clerk's office phone tree uses speech-to-text and text-to-speech but must transfer to a human when recognition confidence is low. Which safeguard should they configure?
Implement information extraction solutions (10–15%) · 27 questions
- A city permits office wants a grounded assistant that can use permitting PDFs, PNG diagrams, and training videos together. What should the retrieval pipeline do first?
- A county public-health bot must answer resident questions by meaning when wording differs from the published guidance titles. Which search approach should they configure for grounding?
- A transit authority needs residents to find both exact route numbers like 42B and natural-language questions about late-night service. Which search approach should they configure?
- A housing authority wants resident policy Q&A grounded in lease embeddings so paraphrased questions still retrieve the right clauses. Which search capability should they configure?
- A clerk’s office is ingesting scanned filing packets that mix text, images, and layout cues. What should the retrieval pipeline apply during indexing?
- A 311 team wants RAG answers about debris-pile complaints that arrive as handwritten photo attachments. What must their ingestion flow include?
- A library wants staff research agents to call the same retrieval pipeline used for the public catalog. How should they expose that pipeline?
- An emergency operations center needs search across audio situation reports and PDF annexes together. What should the index include?
- A water utility replaces its published tariff PDFs and needs the resident bot to stop citing outdated rates. What should the grounding team do to the search index?
- A parks department wants event dates pulled out during enrichment so the search index can filter programs by calendar. What should they add to the ingestion pipeline?
- An open-data portal wants a workflow that drafts dataset descriptions using evidence from related documentation. How should retrieval participate?
- A courts research desk must match exact case citation strings and still support semantic questions about holdings. Why prefer hybrid search over vector-only?
- Public works field agents need RAG answers from handwritten notes on as-built drawing sheets. What should ingestion add to the corpus pipeline?
- The mayor’s office wants press FAQ answers grounded only in approved briefing books, not informal staff drafts. How should retrieval be scoped?
- A clerk must pull structured fields from multimodal permit applications that mix typed text, stamps, and form layout. Which pipeline approach fits?
- A housing authority needs clean grounded document representations of lease PDFs before agents and RAG consume them. Which Foundry capability should they use?
- Transit wants fare-policy PDFs turned into markdown tables that downstream reasoning can consume cleanly. What should their document analyzers output?
- A 311 intake flow must capture checkbox selections and handwritten regions on scanned service forms. What should extraction target?
- Public health receives immunization-clinic packets that mix born-digital PDFs with scanned pages and still needs structured clinic fields. What should the extraction design support?
- A library funding agent needs grant PDF narratives as markdown so it can reason over sections cleanly. What should document analyzers produce?
- Emergency management needs section-aware extraction from thick Emergency Operations Plan PDFs so annexes and chapters stay distinct. Which document capability should they apply?
- A water utility must extract dispute evidence from bill packets that include charts and complex layouts, not just plain paragraphs. How should they choose the extraction approach?
- A courts research agent must not answer from raw unclean OCR alone. What should the document pipeline produce first?
- A parks AP review agent needs structured vendor fields such as amounts and invoice numbers from PDF invoices. What should the extraction step provide?
- An open-data team wants methodology sections from statistical report PDFs turned into structured outputs for cataloging. What should analyzers produce?
- A county assessor needs deed-related scans searchable by extracted fields after OCR. Which pipeline combination should they run?
- A city attorney’s office is building a clause-finding agent over contract PDFs. How should Content Understanding analyzers support that agent?
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.