PMI-CPMAI Study Guide: All 5 Exam Domains and the 6 CPMAI Phases

PMI-CPMAI (PMI Certified Professional in Managing AI) is PMI’s certification for running AI projects with the CPMAI methodology, and you can’t sit the exam until you finish PMI’s 21-hour PMI-CPMAI Exam Prep Course. PMI sells the course and exam as one bundle for $899, or $699 for PMI members (US prices), and the exam is 120 questions in 160 minutes.

This study guide goes through the five domains in PMI’s exam content outline one at a time: what each one lists, how it lines up with the six CPMAI phases taught in the course, and how to study it. If you’re still deciding whether the credential is worth the money, read our honest look at PMI-CPMAI first; this page is about passing it.

PMI-CPMAI at a glance

Checked against PMI’s PMI-CPMAI certification page, the PMI-CPMAI Exam Content Outline (September 2025), the PMI Certification Handbook and the CCR Handbook on October 6, 2026.

What the PMI-CPMAI exam is really testing

PMI built PMI-CPMAI on the CPMAI (Cognitive Project Management in AI) methodology, which it gained when it acquired Cognilytica in September 2024. The exam content outline says the certification focuses on the skills needed to build AI implementations with that methodology, not the general project management knowledge covered by PMI’s other certifications.

In practice, the exam checks whether you can apply CPMAI to manage an AI initiative from inception through operationalization. PMI describes it as vendor-agnostic and covering AI, machine learning, advanced data analytics and intelligent automation projects of any size, with an emphasis on data-driven, iterative work and responsible, trustworthy AI. PMI’s certification page calls the approach tool-agnostic, so you don’t need to know any particular vendor’s AI tools.

Two notes from the outline are worth knowing before you start. The exact number of questions per domain can vary from one exam form to another, and the idea of tailoring your approach to the situation runs through all five domains rather than sitting in one.

Five exam domains, six CPMAI phases

The exam is organized by domain, but the required course is organized by the six CPMAI phases. PMI publishes them separately, so here is how they line up. The mapping is ours, based on the task names in the outline and PMI’s module descriptions.

The weights tell you where to spend time. Domains II and III, the business case and the data, are 52% of the exam between them, and both happen before a model is built. Module 1 of the course, on why AI projects struggle and why iterative delivery helps, sets up the whole method.

Domain I: Support Responsible and Trustworthy AI Efforts (15%)

What the outline lists:

• Oversee the privacy and security plan: data governance for personally identifiable information, encryption and access controls for training data, privacy impact assessments, and compliance with GDPR, CCPA and other data protection rules.
• Manage AI/ML transparency: documenting why a model and data were chosen, transparent reporting on data sources and preprocessing, explainability requirements, audit trails and interpretability tools.
• Conduct bias checks: on the model, the data and the algorithm, including fairness testing across population groups and bias mitigation during development.
• Monitor regulatory and policy compliance: tracking AI regulations and standards, working with legal and compliance teams, and keeping documentation ready for audits.
• Manage accountability documentation and the audit trail: records of development decisions, version control for models, data and training, stakeholder approvals at go/no-go points, and chain of custody for training and test data.

How to study it: because this domain applies in every phase, build a short responsible-AI checklist and add to it as you work through each module: what you document, who approves, and what you check for bias or privacy at that stage. When you practice, be suspicious of any option that skips documentation or approval at a decision point; the outline makes both explicit tasks. For a governance view beyond project management, see our AI governance certifications guide.

Domain II: Identify Business Needs and Solutions (26%)

What the outline lists: ten tasks.

• Identify the problem to be solved, including needs and user personas
• Evaluate initial AI feasibility, including data availability, computing needs, organizational readiness and non-AI alternatives
• Conduct risk assessments covering security, safety and ethics
• Develop the AI project scope statement
• Determine ROI
• Manage adoption and integration risks
• Draft the AI solution (high-level architecture, data flow, model types, integration points)
• Define success criteria such as KPIs and metrics
• Support business case creation
• Identify project resources: people, hardware and contractors

How to study it: practice the first question CPMAI asks of any idea: should this be an AI project at all? The outline’s feasibility task explicitly includes comparing AI against traditional alternatives, and one example under problem identification is mapping business problems to AI patterns. PMI lists its blog post on the seven patterns of AI among the exam’s reference materials, so read it. Then write a one-page scope statement, ROI estimate and set of success criteria for an AI idea from your own job. Our AI readiness checklist is a good source of the organizational readiness questions this domain expects you to ask.

Domain III: Identify Data Needs (26%)

What the outline lists: nine tasks.

• Define the required data: types, formats, volume, granularity and quality criteria
• Identify data subject matter experts, data stewards and governance teams
• Identify data sources and locations, internal and external, including legacy systems
• Coordinate the AI workspace and infrastructure
• Gather the required data and set up refresh procedures
• Check data privacy, compliance, usage rights and access
• Oversee data evaluation: quality, distributions, bias, freshness and exploratory analysis
• Determine whether the data meets the solution’s needs, ending in a go/no-go decision on data readiness
• Convey data understanding to leadership in business terms

How to study it: this domain carries as much weight as the business one. Learn the data quality dimensions (accuracy, completeness, consistency, freshness, representativeness) well enough to spot which one a scenario is describing, and know what evidence supports a data go/no-go. PMI’s reference list includes its post on AI data governance best practices. Our five-question data readiness audit covers the same ground from an IT team’s side.

Domain IV: Manage AI Model Development and Evaluation (16%)

What the outline lists: six tasks.

• Oversee AI/ML model techniques, including the choice between supervised, unsupervised and reinforcement learning and the trade-offs between complexity, performance and interpretability
• Oversee model QA and QC: testing protocols, configuration management for model versions and parameters, peer review
• Manage model training: schedules, resources, hyperparameter tuning, cross-validation and experiment tracking
• Manage data transformation for data preparation: cleaning, feature engineering, normalization, augmentation and synthetic data
• Verify data quality for the go/no-go decision before model training
• Verify the model is ready for operationalization, the final go/no-go before deployment

How to study it: the outline frames every task with “oversee”, “manage” or “verify”, so you need a manager’s grasp of machine learning, not a data scientist’s. Be able to explain the three learning approaches, what cross-validation and hyperparameter tuning are for, and why a more accurate model can be the wrong choice if it can’t be explained. Notice how many go/no-go gates the outline has: data readiness in Domain III, then data quality and model readiness here. Know what evidence each gate needs and who signs off.

Domain V: Operationalize AI Solution (17%)

What the outline lists: seven tasks.

• Manage creation of the deployment plan, including rollback procedures
• Manage the deployment and post-deployment verification
• Oversee model governance: life cycle management, versioning, change control, drift detection and retraining schedules
• Oversee solution metrics, with dashboards and alerts for business and technical KPIs
• Prepare the final report and lessons learned
• Manage the transition plan from the project team to operational support
• Oversee the contingency plan: incident response, backup and disaster recovery, escalation and business continuity

How to study it: the key idea is that an AI project doesn’t end at launch. Models drift as the world changes, so be ready to explain how drift is detected, what triggers retraining and who owns the model after handover. Write a simple transition plan and a contingency plan for an AI system you know. Our guide to building an AI inventory shows how organizations keep track of models once they’re in production.

A study plan for PMI-CPMAI

The PMI-CPMAI Exam Prep Course is mandatory and included in the bundle. PMI describes it as a self-paced, 21-hour course organized around the six CPMAI phases, with scenario-based exercises, case studies, a downloadable workbook and a guided review of the outline’s references. It also earns 21 PDUs toward your other PMI certifications (7 Business Acumen, 11 Ways of Working and 3 Power Skills).

PMI’s outline says the exam isn’t written to any single text and PMI doesn’t endorse specific review courses. Its reference list points to PMI’s own material, including the AI in Project Management page, the free “Introduction: PMI-CPMAI” eLearning, the AI Today podcast, several PMI blog posts and the Leading and Managing AI Projects digital guide. PMI also offers a PMI-CPMAI Practice Exam and a PMI-CPMAI Study Hall if you want extra practice, and offers two publications free to members: the Leading and Managing AI Projects digital guide and The Standard for Artificial Intelligence in Portfolio, Program and Project Management.

A four-week plan (about 33 hours including the course): this pacing is our suggestion. Remember the clock: you have 12 months from purchase to get certified.

In your last week, go through the outline task by task and say out loud what you would do and what you would document. If you can’t, that task is your next study session. Our roundup of the best AI certifications for project managers shows how PMI-CPMAI fits next to other options.

Exam day

• After you finish the course, you provide your exam details and schedule through your myPMI dashboard with Pearson VUE. PMI recommends a test center; online proctored exams need system tests and an extensive check-in.
• There are no scheduled breaks during the 160 minutes. A tutorial before and a survey after can take up to 15 minutes, and that time isn’t counted against you.
• The 20 pretest questions are mixed in randomly and don’t affect your score, so treat every question as if it counts.
• Official results appear on your myPMI dashboard. If you don’t pass, PMI suggests about 30 days of further study before a retake.

Keeping PMI-CPMAI after you pass

PMI-CPMAI runs on a three-year cycle. You need 30 PDUs in that time: at least 18 Education PDUs (including at least 4 each in Ways of Working, Power Skills and Business Acumen) and no more than 12 from Giving Back to the Profession. PDUs can come from learning, teaching, presenting, reading, volunteering and creating content, and renewal also needs a payment based on your region and membership.

Our guide to AI certification renewal rules compares PMI-CPMAI with other AI credentials, and the free renewal tracker helps you keep the date.

PMI-CPMAI vs other AI certifications for managers

PMI-CPMAI costs the most of the four, but the price includes the required course, and it’s the only one built around a project method. If you need a cheaper first AI credential, see our AWS AI Practitioner study guide or Google Generative AI Leader study guide. For the wider field, browse the AI certification index, our guide to generative AI certifications or our ranking of the best AI certifications in 2026. Comparing total costs? The AI certification cost calculator does the math →

HOW TO // AI is not affiliated with or endorsed by PMI. PMI, PMI-CPMAI and CPMAI are marks of Project Management Institute, Inc.; we reference them descriptively. This guide is original and based on PMI’s published exam content outline and certification page. Check the official PMI-CPMAI page for current pricing in your region before you buy.

Keep exploring: AIGP study guide

Related guides

• AI Governance Certifications in 2026: Every Real Option Compared
• AIGP Study Guide: The IAPP AI Governance Exam, Domain by Domain
• The Best AI Certifications in 2026, Ranked: Which Should You Get?
• Generative AI Certifications in 2026: The Real Gen AI Exams Compared