Directory / Microsoft

Microsoft Certified: Azure AI Fundamentals (legacy exam AI-900)

The earlier, purely conceptual version of Azure AI Fundamentals, covering machine learning, vision, language and generative AI at a descriptive level.

What it signals

Practical assessment: Scenario-based multiple choice. Proctored knowledge exam.
Identity assurance: Proctored with identity verification. Was proctored via Pearson VUE; can no longer be taken.
Depth: Foundational. Foundational level.
Experience expected: None. No experience required.
Portability: Mixed. Mostly generic AI concepts, mapped to Azure services.
Verifiability: Public badge or registry. Credential is shareable via a Microsoft Learn verification link/transcript carrying an 'Online Verifiable' tag that an employer can open (profile must be public).

For hiring managers

Cannot be earned any more, but existing holders keep it; fundamentals certifications do not expire and a retired credential stays in the Active section of the transcript for two years after retirement, then moves to Historical.
Conceptual only: no code, no building.

Verifying it

Candidate shares a link from their Microsoft Learn profile: either a per-credential Share link (credentials carry an 'Online Verifiable' tag) or a transcript share link / PDF. The Learn profile must be in public view for the shared credential to be verifiable. Retired credentials remain shareable and stay on the transcript.

What it covers

Published exam outline

Describe Artificial Intelligence workloads and considerations 15-20%
Describe fundamental principles of machine learning on Azure 15-20%
Describe features of computer vision workloads on Azure 15-20%
Describe features of NLP workloads on Azure 15-20%
Describe features of generative AI workloads on Azure 20-25%

By topic

GenAI apps: Substantial (20 to 35%). Generative AI workloads were 20-25% of the final blueprint, at descriptive level.
Agents: Not covered. Agents were not in the AI-900 blueprint.
Classical ML: Some (10 to 20%). ML principles were 15-20%, conceptual only.
Operations: Touched on (under 10%). Only a bullet on model management and deployment capabilities.
Data: Not covered. No data preparation content.
Responsible AI: Some (10 to 20%). Responsible AI principles were about half of a 15-20% domain plus a generative-AI bullet, roughly 10%.

Exam details

Who the provider says it is for. People with technical and non-technical backgrounds wanting to demonstrate knowledge of ML and AI concepts and related Azure services.

Format: Microsoft does not publish per-exam formats. Official question types across the programme: multiple choice, drag and drop, build list, hot area, active screen, case studies, and (on some role-based exams) labs. Exam page states: 'This exam will be proctored. You may have interactive components to complete as part of this exam.'
Delivery: Was proctored via Pearson VUE; can no longer be taken.
Code or hands-on work: No. Purely 'describe/identify' level objectives.
Prerequisites: None
Recommended experience: None; awareness of basic cloud concepts and client-server applications was beneficial. Data science and software engineering experience were not required.
Renewal: None required (fundamentals certifications do not expire)
Price notes: USD figure is from third-party sources; Microsoft prices vary by country.

Preparing

AI-900 study guide (official, archived with retirement banner)

Not confirmed

These details were not available from an official source at the time of research.

priceUsd (official pages only say price depends on country; USD figure is from third-party sources)
questionCount (not published per exam)
exact question formats for this specific exam (Microsoft does not disclose)
online-vs-test-centre delivery options were not confirmed on this exam's own page

Sources (7)