AI-200 — Microsoft Azure AI Cloud Developer Associate Study Plan
A practical AI-200 study schedule for Microsoft Azure AI Cloud Developer Associate candidates, with 7-day, 14-day, 30-day, and 60/90-day paths.
This Study Plan is for candidates preparing for the Microsoft Azure AI Cloud Developer Associate (AI-200) exam from Microsoft. It is designed for working developers and cloud professionals who need a realistic schedule that combines Azure AI concepts, hands-on review, scenario practice, and timed exam readiness.
Use the plan that matches your remaining time. If you are not sure where to start, begin with a diagnostic practice set and use the results to choose your path.
Which plan should you use?
| Time remaining | Use this plan | Best for | Main focus |
|---|---|---|---|
| 7 days | Final review plan | You have already studied or have strong Azure AI experience | Weak-area repair, timed practice, final recall |
| 14 days | Focused plan | You know Azure basics but need structured AI-200 preparation | Fast domain coverage plus mock review |
| 30 days | Balanced plan | You can study most days and want a realistic prep cycle | Learn, practice, review, and test readiness |
| 60/90 days | Full preparation path | You are starting early or need hands-on depth | Complete coverage, labs, spaced repetition, multiple mocks |
A good AI-200 preparation plan should include:
- Diagnostic practice before deep study
- Azure AI service-selection review
- Generative AI and application development practice
- Search, retrieval, grounding, and data workflow review
- Security, identity, monitoring, and troubleshooting review
- Missed-question analysis
- Timed mock exams under realistic conditions
- A final week focused on recall, not new material
AI-200 study map
Organize your study around the skills expected of an Azure AI cloud developer. Do not study services in isolation only; practice choosing the right Azure AI pattern for a scenario.
| Study area | What to review | Practice actions |
|---|---|---|
| Azure AI solution design | Service selection, architecture patterns, integration points | Compare when to use Azure AI services, Azure OpenAI capabilities, search, app hosting, storage, and integration services |
| Generative AI development | Prompts, model interaction, orchestration, grounding, response quality | Build or trace a simple app flow from user request to model response |
| Retrieval and search | Indexing, embeddings, vector search concepts, semantic retrieval, RAG workflow | Explain chunking, indexing, retrieval, grounding, and citation-style response behavior |
| Azure AI service APIs | Language, vision, speech, document processing, and custom AI workflows where relevant | Match scenarios to service capabilities and required inputs/outputs |
| Security and identity | Authentication, authorization, managed identity, secrets, data protection | Review how an application securely calls Azure AI services |
| Monitoring and operations | Logging, metrics, tracing, error handling, latency, cost awareness | Diagnose failed calls, poor responses, access failures, and performance issues |
| Responsible AI | Safety, content filtering, evaluation, bias and harm mitigation, human review | Identify safeguards appropriate for production AI applications |
| Deployment and troubleshooting | SDKs, APIs, configuration, endpoints, network access, integration failures | Practice reading scenario clues and eliminating unsafe or incomplete designs |
Daily practice rhythm
Use the same rhythm most study days. Consistency matters more than long irregular sessions.
If you have 60 minutes
| Minutes | Activity |
|---|---|
| 0-10 | Review yesterday’s missed questions or notes |
| 10-35 | Study one focused topic |
| 35-55 | Complete targeted practice questions |
| 55-60 | Log misses and choose tomorrow’s topic |
If you have 90 minutes
| Minutes | Activity |
|---|---|
| 0-10 | Warm-up: flashcards, service-selection notes, or error patterns |
| 10-40 | Concept review or hands-on walkthrough |
| 40-70 | Practice questions by topic |
| 70-85 | Missed-question review |
| 85-90 | Update weak-area tracker |
If you have 2 to 3 hours
| Block | Activity |
|---|---|
| Block 1 | New topic or hands-on concept review |
| Block 2 | Scenario-based practice questions |
| Block 3 | Review incorrect answers and create short notes |
| Optional | Revisit a previously weak topic for spaced repetition |
Keep a simple tracker with these columns:
| Date | Topic | Questions attempted | Misses | Main reason missed | Retest date |
|---|
Start with a diagnostic
Before choosing what to study, take a diagnostic practice set.
| Step | Action |
|---|---|
| 1 | Take a mixed AI-200 practice set without notes |
| 2 | Mark every question as confident, unsure, or guessed |
| 3 | Review incorrect and guessed questions together |
| 4 | Tag each miss by topic and reason |
| 5 | Build your first weak-area list |
Your diagnostic goal is not to “pass” on day one. It is to find the topics that deserve the most study time.
Common diagnostic tags:
- Service selection
- Azure AI development workflow
- Generative AI application design
- Search, embeddings, and RAG
- Security and identity
- Monitoring and troubleshooting
- Responsible AI
- API, SDK, or configuration detail
- Misread scenario wording
7-day final review plan
Use this plan if the exam is within one week. It assumes you have already studied most objectives or have meaningful Azure AI experience.
Do not try to learn every topic from scratch in the final week. Focus on the highest-value weak areas, timed practice, and recall.
| Day | Main goal | Study actions |
|---|---|---|
| Day 1 | Diagnostic and triage | Take a mixed practice set. Build a ranked list of weak areas. Review every missed question. |
| Day 2 | Azure AI service selection | Review when to use different Azure AI capabilities. Drill scenario questions that ask for the best service, integration pattern, or architecture choice. |
| Day 3 | Generative AI development | Review prompts, grounding, model interaction, evaluation, safety controls, and application flow. Practice scenario questions involving AI app design. |
| Day 4 | Search, retrieval, and data flow | Review embeddings, indexing, retrieval, RAG, data preparation, and response grounding. Practice troubleshooting weak retrieval quality. |
| Day 5 | Security and operations | Review identity, access, secrets, network considerations, monitoring, logging, errors, and cost-aware design. Create a one-page final notes sheet. |
| Day 6 | Timed mock exam | Take a timed mock under exam-like conditions. Spend more time reviewing the mock than taking it. |
| Day 7 | Light final review | Review your notes, missed-question log, and key decision tables. Avoid heavy new material. Prepare logistics and rest. |
Final-week rules
- Stop adding large new topics after Day 5.
- Do not take a full mock late on the night before the exam.
- Review missed questions by reason, not just by correct answer.
- Prioritize scenario interpretation over memorizing isolated facts.
- Keep the final day light and focused.
14-day focused plan
Use this plan if you have two weeks. It is fast, but it gives you enough time for a diagnostic, targeted learning, hands-on review, and timed practice.
| Day | Focus | What to do |
|---|---|---|
| 1 | Diagnostic | Take a mixed practice set. Tag every miss. Build your topic priority list. |
| 2 | Exam map and Azure AI architecture | Review the latest Microsoft AI-200 skills outline and map your weak areas to major study topics. |
| 3 | Azure AI service selection | Practice matching scenarios to services, APIs, and architecture patterns. |
| 4 | Azure AI services for content understanding | Review language, vision, speech, and document-style workflows as relevant to AI app development. |
| 5 | Generative AI app flow | Review prompts, model calls, orchestration, input/output handling, and response validation. |
| 6 | Grounding and RAG | Study embeddings, indexing, retrieval, chunking concepts, and grounded response patterns. |
| 7 | Mini mock and review | Take a timed half-length or large mixed set. Review misses deeply. |
| 8 | Security and identity | Review authentication, authorization, managed identity, secrets, and secure app-to-service communication. |
| 9 | Monitoring and troubleshooting | Review logging, metrics, tracing, errors, latency, failed API calls, and poor AI output quality. |
| 10 | Responsible AI | Review safety, content controls, evaluation, human review, and production safeguards. |
| 11 | Deployment and integration | Review endpoints, SDK/API usage, configuration, app hosting, data connections, and integration troubleshooting. |
| 12 | Full timed mock | Take a full timed mock. Rebuild your weak-area list from the results. |
| 13 | Weak-area sprint | Drill the top 3 weak areas. Reattempt missed questions from Days 1-12. |
| 14 | Final review | Light review only. Use notes, flashcards, service-selection tables, and error-pattern review. |
Stop adding new material after Day 12 unless the topic is a clear blocker. Days 13 and 14 are for consolidation.
30-day balanced plan
Use this plan if you have about one month. It provides enough time to learn, practice, revisit, and test.
Weekly structure
| Week | Goal | Outcome by end of week |
|---|---|---|
| Week 1 | Baseline and core Azure AI coverage | You know your weak areas and can explain major Azure AI service choices |
| Week 2 | Generative AI and retrieval workflows | You can reason through AI app design, grounding, search, and evaluation scenarios |
| Week 3 | Security, operations, deployment, troubleshooting | You can handle production-style scenario questions |
| Week 4 | Mock exams and weak-area repair | You are using timed practice and reviewing misses efficiently |
Day-by-day schedule
| Days | Focus | Study actions |
|---|---|---|
| 1 | Diagnostic | Take a mixed practice set. Tag misses. Build your tracker. |
| 2-3 | Exam objectives and architecture | Review the Microsoft AI-200 skills outline. Create a checklist of topics to cover. |
| 4-5 | Azure AI service selection | Practice scenario questions where multiple services appear plausible. |
| 6-7 | Azure AI service APIs | Review common AI service workflows, inputs, outputs, authentication, and integration points. |
| 8-10 | Generative AI development | Study prompt patterns, model calls, response handling, grounding, and evaluation. |
| 11-13 | Retrieval and search | Review search indexes, embeddings, vector retrieval concepts, chunking, and RAG workflow design. |
| 14 | Mixed review set | Take a timed mixed set. Update weak-area ranking. |
| 15-16 | Security and identity | Review secure application access, managed identity, secrets, permissions, and data protection. |
| 17-18 | Monitoring and troubleshooting | Practice diagnosing access errors, failed calls, poor relevance, latency, and configuration issues. |
| 19-20 | Responsible AI and safety | Review safety controls, evaluation, content handling, human oversight, and production guardrails. |
| 21 | Mock exam 1 | Take a timed mock. Review all misses and guesses. |
| 22-24 | Weak-area repair | Re-study your lowest-scoring topics. Use targeted question sets. |
| 25 | Hands-on concept review | Walk through an end-to-end AI app architecture: data source, retrieval, model call, security, monitoring. |
| 26 | Scenario drill day | Focus on long scenario questions and “best option” decision-making. |
| 27 | Mock exam 2 | Take another timed mock under realistic conditions. |
| 28 | Mock review | Review the mock deeply. Rewrite your final notes sheet. |
| 29 | Final weak-area sprint | Reattempt missed questions and review service-selection notes. |
| 30 | Light review | No heavy new material. Review notes, rest, and prepare exam logistics. |
30-day pacing tips
- Study new material mostly in the first 3 weeks.
- Use Week 4 for timed practice and review.
- Do not measure readiness from one mock score only; look for repeated improvement.
- If you miss the same type of question twice, do a focused mini-review before attempting more questions.
60/90-day full preparation path
Use this path if you are starting early, changing roles, or need hands-on depth with Azure AI development.
| Phase | 60-day timing | 90-day timing | Focus |
|---|---|---|---|
| Phase 1: Baseline | Days 1-5 | Days 1-7 | Diagnostic, exam map, study tracker, environment setup |
| Phase 2: Core Azure AI | Days 6-18 | Days 8-25 | Azure AI services, service selection, APIs, app integration |
| Phase 3: Generative AI apps | Days 19-30 | Days 26-45 | Prompts, model calls, orchestration, grounding, evaluation |
| Phase 4: Search and data | Days 31-40 | Days 46-60 | Retrieval, embeddings, search, indexing, RAG workflows |
| Phase 5: Secure production design | Days 41-48 | Days 61-72 | Identity, secrets, network access, monitoring, troubleshooting |
| Phase 6: Mock and repair | Days 49-56 | Days 73-84 | Timed mocks, weak-area repair, scenario drills |
| Phase 7: Final review | Days 57-60 | Days 85-90 | Final notes, reattempt misses, light review, exam readiness |
60-day weekly plan
| Week | Focus | Practice target |
|---|---|---|
| 1 | Diagnostic and AI-200 objective mapping | One mixed diagnostic set |
| 2 | Azure AI services and service-selection decisions | Topic drills by service category |
| 3 | AI app integration and APIs | Scenario questions plus hands-on walkthroughs |
| 4 | Generative AI development | Prompt, grounding, response validation, and safety questions |
| 5 | Retrieval, search, embeddings, and RAG | Architecture and troubleshooting drills |
| 6 | Security, identity, monitoring, and operations | Production-readiness scenario sets |
| 7 | Mock exam 1 and weak-area repair | Full timed mock plus review |
| 8 | Mock exam 2 and final review | Final timed mock, notes, and reattempts |
90-day weekly plan
| Weeks | Focus | What to add compared with the 60-day plan |
|---|---|---|
| 1-2 | Baseline and Azure AI fundamentals | More time with official objective mapping and concept notes |
| 3-4 | Azure AI service workflows | Deeper hands-on review of service inputs, outputs, and integration patterns |
| 5-6 | Generative AI application development | More practice with prompt design, grounding, evaluation, and responsible AI |
| 7-8 | Search, retrieval, and data preparation | More scenario work around relevance, indexing, chunking, and retrieval failures |
| 9-10 | Security and operations | More review of identity, access, monitoring, error handling, and deployment concerns |
| 11 | Mock exam and weak-area repair | First full timed mock and targeted remediation |
| 12 | Final mock and review | Second full timed mock, final notes, and retesting |
| 13 | Final exam readiness | Light review, confidence checks, and exam logistics |
Domain-by-domain drill plan
Use this table to convert broad topics into concrete study sessions.
| Domain-style topic | Study questions to answer | Drill activity |
|---|---|---|
| Azure AI service selection | Which service or capability best fits the scenario? What are the tradeoffs? | Create a comparison table of common use cases and the Azure AI capabilities that match them |
| Generative AI app design | How does the app pass user input, retrieve context, call the model, and handle output? | Draw the flow from user request to response, including validation and logging |
| Prompt and response handling | What instructions, context, examples, or constraints are needed? | Rewrite poor prompts into clearer task-specific prompts |
| RAG and search | How is content prepared, indexed, retrieved, and used for grounding? | Explain chunking, embeddings, retrieval, ranking, and answer generation in sequence |
| Security | How does the app authenticate and protect secrets and data? | Review managed identity, RBAC-style access, secret storage, and least privilege concepts |
| Monitoring | How do you know the app is failing, slow, expensive, or producing weak answers? | List the logs, metrics, traces, and review signals you would check |
| Troubleshooting | What clue points to configuration, identity, network, data quality, or model behavior? | Convert each missed troubleshooting question into a root-cause note |
| Responsible AI | What safeguards are needed before production use? | Identify safety filters, evaluation steps, user disclosures, and human review points |
Hands-on concept review
Even if the exam is multiple choice, AI-200 preparation should include practical architecture thinking. You do not need to overbuild labs, but you should be able to explain how a real Azure AI application works.
Use these hands-on review prompts:
| Scenario | What to practice explaining |
|---|---|
| A web app calls an Azure AI service | Authentication, endpoint configuration, input validation, error handling, and logging |
| A generative AI app uses company documents | Data ingestion, chunking, indexing, retrieval, grounding, and response evaluation |
| An AI response is inaccurate | Prompt quality, missing context, retrieval quality, evaluation, and human review |
| An API call fails | Credentials, permissions, endpoint, network path, request format, quota or throttling symptoms without relying on exact limits |
| Search results are poor | Content quality, indexing strategy, query design, embeddings, filters, and ranking |
| A production app needs safeguards | Content safety, responsible AI review, monitoring, auditability, and fallback behavior |
For each hands-on review, write a short answer to three questions:
- What is the intended architecture?
- What can fail?
- How would I diagnose or improve it?
Missed-question review method
Do not just read the explanation and move on. Most score improvement comes from reviewing misses correctly.
Classify every miss
| Miss type | What it means | Fix |
|---|---|---|
| Knowledge gap | You did not know the concept | Re-study the topic and create a short note |
| Service-selection error | You chose a plausible but wrong Azure service or capability | Build a comparison table and drill similar scenarios |
| Scenario misread | You missed a key requirement or constraint | Underline requirements before choosing an answer |
| Security blind spot | You ignored identity, access, secrets, or data protection | Review secure design patterns |
| Troubleshooting error | You treated a symptom as the root cause | Write the likely root cause and first diagnostic step |
| Overthinking | You rejected the direct answer without evidence | Practice choosing the option that satisfies the stated requirement |
| Time pressure | You rushed or spent too long | Use timed sets and a skip/return strategy |
Review sequence
For each missed or guessed question:
- Write the topic tag.
- Write why your answer was wrong.
- Write the rule or decision point in your own words.
- Identify the clue in the question that should have guided you.
- Create one mini-drill or flashcard.
- Reattempt a similar question within 2 to 4 days.
A good missed-question note is short:
“Missed because I treated this as a model issue, but the scenario described poor retrieval context. First check indexing, chunking, query, and grounding before changing the model.”
When to use timed mock exams
Timed mocks are for readiness and pacing. Topic drills are for learning. Use both, but do not replace study with endless mocks.
| Plan | First timed mock | Second timed mock | Final mock |
|---|---|---|---|
| 7 days | Day 1 diagnostic or Day 6 full mock | Optional only if stamina is weak | Day 6, not the night before |
| 14 days | Day 7 mini mock | Day 12 full mock | Day 12 or 13 |
| 30 days | Day 21 | Day 27 | Day 27, then review |
| 60 days | Around Week 7 | Around Week 8 | Several days before exam |
| 90 days | Around Week 11 | Around Week 12 | Several days before exam |
Mock exam rules:
- Take full mocks without notes.
- Use a timer.
- Mark uncertain questions.
- Review both incorrect and guessed questions.
- Do not memorize practice questions; study the reasoning.
- Treat any score as a readiness signal, not an official pass prediction.
Exam-readiness checks
Use these checks in the final week.
| Readiness area | You are ready when… |
|---|---|
| Service selection | You can explain why the correct Azure AI option fits better than the distractors |
| Scenario reading | You identify requirements, constraints, and security clues before answering |
| Generative AI | You can reason through prompts, grounding, model calls, output validation, and safety |
| Retrieval and search | You can troubleshoot poor relevance, missing context, and weak grounded answers |
| Security | You consistently consider identity, least privilege, secrets, and data protection |
| Operations | You know what to check for failed calls, latency, monitoring gaps, and configuration issues |
| Responsible AI | You can identify appropriate safeguards for production AI applications |
| Timing | You can complete timed sets without rushing the final questions |
| Review discipline | You can explain your missed questions without simply memorizing answers |
As a personal readiness target, look for consistent performance on fresh practice questions, fewer repeated mistakes, and clear explanations for why each correct answer is correct. Do not rely on one mock score alone.
Final 48 hours
| Time | What to do | What to avoid |
|---|---|---|
| 48 hours out | Review weak-area notes and reattempt selected misses | Starting a brand-new large topic |
| 24 hours out | Review service-selection notes, security reminders, and troubleshooting patterns | Taking an exhausting full mock late in the day |
| Exam day | Warm up lightly with notes or a few easy questions | Cramming unfamiliar material |
Final checklist:
- Confirm exam appointment details.
- Prepare identification and testing environment requirements.
- Sleep normally if possible.
- Use a steady question pace.
- Flag difficult questions and return later.
- Read scenario constraints before looking for the answer.
Practical next step
Start with one mixed AI-200 diagnostic practice set. Tag every missed or guessed question, choose the plan that matches your remaining time, and make your next study session a targeted review of your weakest Azure AI topic.