Your comprehensive guide to the refreshed AI-901 certification, built around Microsoft Foundry. Covers responsible AI, generative AI, agents, multimodal experiences, and Azure Content Understanding for candidates at all experience levels.
Developers, career changers, students, and business professionals learning to build AI solutions with Microsoft Foundry.
40-60 multiple-choice questions in 45 minutes; passing score is 700/1000. No prerequisites required.
Responsible AI principles, generative AI and agents in Foundry, multimodal vision and speech apps, and Azure Content Understanding for information extraction.
AI-900 retires on 30 June 2026. AI-901 pivots from individual services to the unified Microsoft Foundry platform, with new coverage of agents and Content Understanding.
If you are currently studying for AI-900, here is what you need to know:
The Microsoft AI-901 exam, officially titled Azure AI Fundamentals (Refreshed), is the modern entry point for anyone looking to validate their knowledge of artificial intelligence on the Microsoft cloud. It replaces the retiring AI-900 exam and has been rebuilt around Microsoft Foundry, the unified platform for building, deploying, and managing AI solutions on Azure.
This certification validates your foundational knowledge of AI concepts and responsibilities, and your practical ability to implement generative AI apps, agents, multimodal experiences, and information extraction solutions using Foundry. It is designed for both technical and non-technical candidates, and whilst it is not a prerequisite for higher-level certifications like Azure AI Engineer Associate, it provides a strong foundation for those pursuing them.
Individuals looking to pivot into the AI field and need a recognised credential that demonstrates hands-on Foundry skills.
Managers, product owners, and analysts who need to understand how AI solutions are built in Foundry to make informed decisions.
Academic learners wanting to supplement their degree with an industry-standard certification in generative AI and agent development.
Python-familiar developers who want a structured introduction to building chat clients, agents, and multimodal apps with the Foundry SDK.
The refreshed AI-901 exam covers two main sections. Section 2 carries the majority of the exam weight, reflecting the shift toward practical implementation of AI solutions in Microsoft Foundry.
AI-901 is built around Microsoft Foundry as the unified platform for building, deploying, and managing AI solutions. Focus your study on these four areas.
The unified workspace for deploying models, testing prompts, building agents, and running lightweight client applications. The Foundry SDK is the primary developer surface.
Create, test, and deploy single-agent solutions directly in the Foundry portal. Candidates must build a lightweight client application that interacts with an agent.
Foundry Tools service for extracting structured information from documents, forms, images, audio, and video. A centrepiece of the information extraction section.
Deploy multimodal models that handle vision, speech, and image generation in a single configuration, replacing the AI-900 pattern of separate Vision and Language services.
Test your knowledge with 200+ exam-realistic questions built for the refreshed April 2026 objectives. Get detailed explanations and track your progress.
This three-week plan mirrors the official Section 1 and Section 2 weightings, with two of the three weeks dedicated to implementing AI solutions in Microsoft Foundry.
Foundations and responsible AI
Responsible AI principles: fairness, reliability, safety, privacy, inclusiveness, transparency, and accountability.
How generative AI models work, choosing an appropriate model by capability, and model deployment configuration parameters.
Common AI workloads (generative, agentic, text, speech, vision), text analysis techniques, and information extraction patterns.
Building with Microsoft Foundry
Foundry portal orientation, model deployment, writing effective system and user prompts, and interacting with deployed models.
Lightweight chat clients with the Foundry SDK, creating and testing single-agent solutions, and client apps for agents.
Text analysis apps, multimodal models for spoken prompts, and Azure Speech in Foundry Tools.
Multimodal and information extraction
Interpreting visual input with multimodal models, image generation, and lightweight vision-enabled applications.
Azure Content Understanding for documents, forms, images, audio, and video, plus building information extraction applications.
Full-length practice exams, targeted review of weak areas, and final exam preparation.