AZ-204 停考後的 AI-200 Azure AI Cloud Developer 考試指南
AZ-204 Retired: AI-200 Study Guide for Azure AI Cloud Developers
作者整理 AI-200 備考指南，説明 AZ-204 已於 2026 年 7 月 31 日停考，而 AI-200 是現行中級 Azure 開發者考試，兩者考綱不同。指南列出 AI-200 考試為 120 分鐘、700 分合格，並比較考綱比重及兩項考試的內容差異。另提供六週、每週約六小時的備考計劃，以及容器、數據管理服務和訊息處理等實作練習。
Originally published on EpiFive, where it is kept up to date.
AZ-204, the Azure Developer Associate exam, retired on 31 July 2026. The current intermediate Azure developer exam is AI-200, Developing AI Cloud Solutions on Azure, which earns the Microsoft Certified: Azure AI Cloud Developer Associate certification. Microsoft's pages do not call it a like-for-like replacement, and its outline is different. It is a 120-minute proctored exam with a pass mark of 700, and its outline shifts from general Azure development to containers, vector search in databases, messaging, serverless and observability. This guide compares the two outlines and gives a study plan for AI-200.
Who this is for
This is for developers who searched for an AZ-204 study guide, for people who planned to take AZ-204 and missed the date, and for anyone choosing an intermediate Azure developer certification. At the end you will know what AI-200 measures, how the weights should shape your time, which hands-on tasks to practice, and what to do if you hold AZ-204.
The information comes from Microsoft Learn pages checked on 26 October 2026: the AZ-204 study guide and certification page, the AI-200 certification page and the AI-200 study guide (skills measured as of 7 October 2026). Outlines change, so reread the study guide before you book.
What happened to AZ-204
The AZ-204 study guide carries this warning: the exam was retired on 31 July 2026 at 11:59 PM Central Standard Time. The Azure Developer Associate certification page warns that the certification and the renewal assessment are retired. Microsoft's last AZ-204 outline (skills measured as of 14 January 2026) had five areas:
| AZ-204 skill area | Weight |
|---|---|
| Develop Azure compute solutions | 25-30% |
| Develop for Azure storage | 15-20% |
| Implement Azure security | 15-20% |
| Monitor, troubleshoot and optimize Azure solutions | 5-10% |
| Connect to and consume Azure services and third-party services | 20-25% |
Microsoft's pages do not map AZ-204 holders to AI-200 automatically. If you hold the certification, check your Microsoft Learn transcript for its status, and do not assume a renewal is available.
What AI-200 is
Microsoft describes the certification as validating your ability to design, build and implement AI solutions on Azure, with a focus on back-end services, scalable architectures and the full development lifecycle. The audience profile says you contribute to all phases of implementing AI solutions on Azure, with emphasis on back-end services and components, and that you should be proficient in:
- Azure SDKs and third-party SDKs used in Azure.
- Azure data management services.
- Azure monitoring and troubleshooting.
- Azure messaging and eventing.
- Vector databases.
- Python programming.
- Implementing containerized applications on Azure.
Exam facts
| Item | Value from Microsoft |
|---|---|
| Exam | AI-200: Developing AI Cloud Solutions on Azure |
| Certification | Microsoft Certified: Azure AI Cloud Developer Associate |
| Level | Intermediate |
| Length | 120 minutes |
| Passing score | 700 or greater |
| Format | Proctored, may include interactive components |
| Languages | English, Arabic (Saudi Arabia), Chinese (Simplified), Chinese (Traditional), French, German, Indonesian (Indonesia), Italian, Japanese, Korean, Portuguese (Brazil), Russian, Spanish |
| Price | Based on the country or region in which the exam is proctored |
| Practice assessment | On AI Skills Navigator, and you must be signed in to launch it |
| Skills outline | As of 7 October 2026 |
Microsoft strongly recommends registering with a personal Microsoft account, because records tied to an organizational account are lost if you leave the organization. Associate certifications expire annually and can be renewed with a free online assessment, per Microsoft's study guide.
Skills measured
| Skill area | Weight in the "at a glance" list |
|---|---|
| Develop containerized solutions on Azure | 20-25% |
| Develop AI solutions by using Azure data management services | 30-35% |
| Connect to and consume Azure services | 20-25% |
| Secure, monitor and troubleshoot Azure solutions | 20-25% |
Microsoft's study guide shows 30-35% for the data management area in its summary list but 25-30% in that area's own heading. The change log notes the percentage of the exam increased for that area. Treat it as the largest area and check the study guide for the final figure.
Containerized solutions (20-25%)
- Container hosting: build, store, version and manage images with Azure Container Registry; build and run images with Azure Container Registry Tasks; deploy containers to Azure App Service, including configuring App Service to supply environment variables and secrets.
- Orchestration: deploy to Azure Container Apps, including environment configuration and revision management; implement event-driven scaling with Kubernetes Event-driven Autoscaling (KEDA) in Container Apps; deploy and manage applications on Azure Kubernetes Service by using manifest files; monitor and troubleshoot AKS and Container Apps by inspecting logs, events and end-to-end connectivity.
Practice: write a Dockerfile for a small Python API, push it to a registry, run it on Container Apps, add a scale rule, and deploy the same image to AKS with a manifest.
AI solutions with data management services (largest area)
- Azure Cosmos DB for NoSQL: connect with the SDK and run queries; optimize query performance and Request Unit consumption through indexing policies and consistency levels; store and retrieve embeddings and run vector similarity search; implement a change feed processor.
- Azure Database for PostgreSQL: connect and query with SDKs; model schemas and choose data types; implement indexing strategies, including optimizing query latency and reducing pgvector compute overhead; configure compute, memory and storage for vector workloads; run vector similarity search, including storing embeddings, semantic retrieval and retrieval-augmented generation (RAG) patterns with metadata filters; implement connection optimization.
- Azure Managed Redis: caching, expiration and invalidation; vector indexing for similarity search.
Practice: store embeddings in Cosmos DB and in PostgreSQL with pgvector, run a similarity query with a metadata filter, and compare indexing choices. For the PostgreSQL side, see pgvector in PostgreSQL: HNSW and IVFFlat tuning, and for reading query plans, PostgreSQL EXPLAIN ANALYZE.
Connect to and consume Azure services (20-25%)
- Event- and message-based: queue and process back-end operations with Azure Service Bus, including dead-letter queue handling, messages, topics and subscriptions; build event-driven workflows with Azure Event Grid, including filters, custom events and retries.
- Azure Functions: build serverless APIs with triggers and bindings; configure and deploy function apps.
Practice: send messages to a Service Bus topic with two subscriptions, force a message to the dead-letter queue and read it, route a custom Event Grid event with a filter, and build an HTTP-triggered function with a queue output binding.
Secure, monitor and troubleshoot (20-25%)
- Secure solutions: secure secrets with Azure Key Vault, including rotation and retrieval; store and retrieve configuration with Azure App Configuration.
- Monitor and troubleshoot: trace distributed systems with OpenTelemetry SDKs; write KQL queries to analyze logs and metrics.
Practice: keep a secret in Key Vault and read it from an app using a managed identity (the outline's wording for this area names Key Vault and App Configuration, so check the study guide for identity-specific objectives), instrument a service with OpenTelemetry, and write KQL to find failed requests. For KQL practice, see Advanced hunting KQL for identity attacks, which uses the same query language in another product.
How AI-200 differs from AZ-204
Comparing the two published outlines:
| Topic | AZ-204 outline | AI-200 outline |
|---|---|---|
| Containers | Images, Container Registry, Container Instances, Container Apps | Container Registry and Tasks, App Service containers, Container Apps with KEDA, AKS manifests |
| App Service Web Apps | Create, diagnostics, deploy, TLS and settings, autoscale, slots | Appears only as a container deployment target |
| Functions | Create, bindings, triggers | Serverless APIs, triggers, bindings, configure and deploy |
| Cosmos DB | SDK operations, consistency, change feed | SDK, queries, indexing and RU optimization, vector search, change feed processor |
| Blob Storage | Properties, SDK, lifecycle policies | Not in the outline |
| PostgreSQL and Redis | Not in the outline | Azure Database for PostgreSQL with pgvector, Azure Managed Redis |
| Identity | Microsoft Identity platform, Entra ID, SAS, Graph, managed identities | Key Vault and App Configuration only |
| API Management | Create, document, configure, policies | Not in the outline |
| Events and messages | Event Grid, Event Hubs, Service Bus, Queue Storage | Service Bus and Event Grid |
| Monitoring | Application Insights | OpenTelemetry and KQL |
This is a comparison of the published outlines only. "Not in the outline" means the skill is not listed in the AI-200 study guide, though Microsoft notes that related topics may be covered.
A six-week study plan
This plan assumes about six hours a week and some existing experience with Azure and Python.
| Week | Focus | Output |
|---|---|---|
| 1 | Containers: Registry, Tasks, App Service | An image built with ACR Tasks and running on App Service with secrets supplied as settings |
| 2 | Container Apps with KEDA, AKS manifests | A scaled Container App and an AKS deployment, with logs inspected |
| 3 | Cosmos DB: SDK, indexing, RUs, vectors, change feed | A small app that stores embeddings and runs similarity search |
| 4 | PostgreSQL with pgvector, Azure Managed Redis | Vector search with a metadata filter, plus a cache with expiry and invalidation |
| 5 | Service Bus, Event Grid, Functions | A message flow with a dead-letter path, a filtered event and a function app |
| 6 | Key Vault, App Configuration, OpenTelemetry, KQL, practice assessment | Traces in Azure Monitor, a KQL query, and a gap list from the practice assessment |
Build in a real subscription
The outline is written as tasks, so you need a subscription to practice. Set a budget and an alert, use the lowest tiers that support each feature, and delete resource groups when you finish a lab. Some services in the outline, such as AKS and managed databases, can incur cost while idle.
Use Microsoft's resources
The study guide links to Azure documentation for each service, Microsoft Q&A, the Exam Readiness Zone videos and the exam sandbox. Its "Get trained" link still points at the AZ-204 preparation page, so if a link looks stale, start from the AI-200 certification page instead.
Troubleshooting study problems
| Problem | What to do |
|---|---|
| Old AZ-204 notes cover Blob Storage, API Management and Entra authentication | Keep them for background, but rebuild your plan from the AI-200 outline |
| You have no AI background | Focus on the data and messaging objectives first. Vector search is a storage and indexing task: embeddings in, similarity query out |
| Practice assessment link does not open | Microsoft says it is on AI Skills Navigator and requires sign-in |
| Costs rise during labs | Use budgets and alerts, and delete labs after each session |
Checklist
- Confirm you are studying AI-200, not the retired AZ-204.
- Read the study guide and note the skills measured date (7 October 2026 at the time of writing).
- Practice each task in a real subscription with cost controls.
- Spend the most time on Cosmos DB, PostgreSQL and Redis with vectors.
- Take the practice assessment on AI Skills Navigator and map misses to outline bullets.
- Book the 120-minute proctored exam with a personal Microsoft account.
For a later step in the Azure path, see AZ-400 study guide: DevOps Engineer Expert.
References
- Exam AI-200 and Azure AI Cloud Developer Associate: https://learn.microsoft.com/en-us/credentials/certifications/azure-ai-cloud-developer-associate/
- Study guide for Exam AI-200: https://learn.microsoft.com/en-us/credentials/certifications/resources/study-guides/ai-200
- Study guide for Exam AZ-204 (retired): https://learn.microsoft.com/en-us/credentials/certifications/resources/study-guides/az-204
- Azure Developer Associate certification page (retired): https://learn.microsoft.com/en-us/credentials/certifications/azure-developer/
This guide was researched and drafted with AI assistance and checked against the vendor documentation listed in its References. Read the maintained version and related guides on EpiFive.
來源：DEV Community · dev.to