Professional Scrum Master - AI Essentials (PSM-AI) is Scrum.org's certification for Scrum Masters who use AI in their work. It does not test whether you can build a model. It tests whether you understand what AI tools can and cannot do, how to use them well and responsibly, and how they fit into Scrum without replacing it.
The exam at a glance
- Questions: 40, multiple choice, multiple answer and true or false
- Time: 60 minutes, about 90 seconds per question
- Pass mark: 85%, so you can miss at most 6 questions
- Cost: $200 USD per attempt
- Prerequisite: none; Scrum.org recommends its AI Essentials class but does not require it
- Validity: the certification does not expire
The time is more generous than in PSM I, where you have 80 questions in 60 minutes. The difficulty lies elsewhere: many questions describe a situation and ask what a Scrum Master should do, and several answers sound reasonable.
The four focus areas
1. AI Theory and Primer
How large language models produce answers, why they hallucinate, and what retrieval-augmented generation (RAG), fine-tuning and AI agents are. You need the concepts, not the mathematics. A good mental model: a language model predicts likely next words, so a fluent answer is not the same as a true answer.
2. Effective AI Prompting
What makes a good prompt: a clear task, the right context, examples, constraints and the format you want back. Expect questions on iterating, on asking the model to ask clarifying questions, and on reviewing what comes back before you use it.
3. AI Security and Ethics
Data protection and approved tools, bias and fairness, accountability and transparency, and the principles of trustworthy AI from the EU and the OECD. Prompt injection and deepfakes also come up. The recurring theme: people stay accountable for what they do with AI output.
4. AI for Scrum Masters
How AI can support Sprint Planning, the Daily Scrum, the Sprint Review and the Sprint Retrospective without taking over the conversations those events exist for. What AI-generated work means for the Definition of Done, how to introduce AI empirically, and how a Scrum Master coaches a team and an organisation on its use.
Video Blocked for Privacy
This video is hosted on YouTube. Loading it will transfer data to Google.
The video above, by 3Blue1Brown, is one of the videos in our first lesson. In a few minutes it shows how a language model turns text into tokens and predicts the next one, which explains most of what the exam asks about hallucinations.
A typical question
A Developer uses an AI assistant to write the code for a Product Backlog item. The code passes all tests. Who is accountable for its quality?
The answer is the Developers. AI does not change the accountabilities in Scrum. Work created with AI help has to meet the Definition of Done like any other work, and the Developers remain accountable for the quality of the Increment. Answers that shift accountability to the tool, to its vendor or to the Scrum Master are traps.
Five tips for the exam
- Think Scrum first. Most AI for Scrum Masters questions are Scrum questions in disguise. If an answer would weaken an accountability, an event or empiricism, it is probably wrong.
- AI supports, people decide. Answers in which AI makes the decision, for example on the Sprint Goal, are rarely right.
- Read for the qualifier. Words like "always", "never" and "only" often mark the wrong option.
- Count the answers. Multiple answer questions tell you how many to pick. Read every option before you choose.
- Practise under time. Ninety seconds per question is enough, but only if you do not get stuck. Mark hard questions and come back to them.
A study plan that works
Our PSM-AI Essentials learning path follows exactly these four focus areas. You can set it to 24 hours, one week or longer, and it hides the optional steps when time is short.
- Learn: one lesson per focus area, with videos, a narrated version, the text and a quiz.
- Practise: Scrum.org's free Scrum Master AI Open, then our two practice exams in the real format.
- Pass: the exam tips, then book the assessment once you score well above 85% in practice.
All details of the certification are on our PSM-AI Essentials certification page, and you can try 15 questions from the first practice exam for free.