PMI Certified Professional in Managing AI (PMI-CPMAI) Exam Guide 2026: Domains, Format, and How to Prepare

What the PMI-CPMAI exam covers

The PMI-CPMAI exam is built from the PMI-CPMAI Examination Content Outline and Specifications (September 2025), which organizes the exam into five domains. The exam itself consists of 120 questions (100 scored and 20 unscored pretest questions) administered over 160 minutes. Each domain carries a specific, fixed weight toward the scored portion of the exam.

Support Responsible and Trustworthy AI Efforts (15%)

This domain focuses on governance and ethics surrounding AI systems. Tasks include:

• ECO-I-T1 Oversee privacy and security plan
• ECO-I-T2 Manage AI/ML transparency (e.g., data selection, algorithm selection)
• ECO-I-T3 Conduct bias checks (e.g., model, data, algorithm)
• ECO-I-T4 Monitor regulatory and policy compliance
• ECO-I-T5 Manage accountability documentation and audit trail

Identify Business Needs and Solutions (26%)

Tied with Identify Data Needs as the heaviest-weighted domain, this covers the front end of an AI initiative — defining the problem, assessing feasibility, and building the business case. Tasks include:

• ECO-II-T1 Identify problem to be solved (e.g., needs, persona)
• ECO-II-T2 Evaluate initial AI feasibility
• ECO-II-T3 Conduct risk assessment(s) (e.g., security, safety, ethics)
• ECO-II-T4 Develop AI project scope statement
• ECO-II-T5 Determine ROI
• ECO-II-T6 Manage adoption/integration risks
• ECO-II-T7 Draft AI solution
• ECO-II-T8 Define success criteria (e.g., KPIs, metrics)
• ECO-II-T9 Support business case creation
• ECO-II-T10 Identify project resources (e.g., people, hardware, contractors)

Identify Data Needs (26%)

Also weighted at 26%, this domain is about sourcing, evaluating, and validating the data that will feed an AI solution. Tasks include:

• ECO-III-T1 Define required data
• ECO-III-T2 Identify data SMEs
• ECO-III-T3 Identify data sources and locations
• ECO-III-T4 Coordinate AI workspace and infrastructure
• ECO-III-T5 Gather required data
• ECO-III-T6 Check data privacy, compliance, and access
• ECO-III-T7 Oversee data evaluation
• ECO-III-T8 Determine if data meets solution needs
• ECO-III-T9 Convey data understanding to leadership

Manage AI Model Development and Evaluation (16%)

This domain addresses the technical oversight of building and validating a model. Tasks include:

• ECO-IV-T1 Oversee AI/ML model technique(s) (e.g., algorithm, selection)
• ECO-IV-T2 Oversee AI/ML model QA/QC (e.g., configuration management, model performance)
• ECO-IV-T3 Manage AI/ML model training
• ECO-IV-T4 Manage data transformation to conduct data preparation
• ECO-IV-T5 Verify data quality for go/no-go decision to conduct data preparation
• ECO-IV-T6 Verify model ready for operationalization go/no-go decision

Operationalize AI Solution (17%)

This domain covers deployment, governance, and lifecycle management after a model is built. Tasks include:

• ECO-V-T1 Manage creation of AI solution deployment plan
• ECO-V-T2 Manage AI solution deployment
• ECO-V-T3 Oversee model governance
• ECO-V-T4 Oversee AI solution metrics (e.g., KPI, model performance)
• ECO-V-T5 Prepare final report/lessons learned
• ECO-V-T6 Manage AI solution transition plan
• ECO-V-T7 Oversee AI solution contingency plan

How the exam is scored

PMI-CPMAI uses a band scoring model. Rather than publishing a single raw numeric score, results are reported in performance bands that reflect how a candidate performed relative to the scored content. Only the 100 scored questions count toward the result; the 20 unscored pretest questions are mixed in for future exam development and do not affect the outcome. Because the exact passing threshold and the scaling behind the bands are determined by PMI, candidates should not rely on any third-party source — including this guide — for a specific passing score. For authoritative scoring detail, consult PMI's own documentation at the official PMI-CPMAI content outline page.

How much to study

There is no official study-hour requirement, and any number offered here is an estimate based on the structure of the content outline, not a documented fact. A reasonable way to plan is to weight your study time roughly in proportion to each domain's share of the exam:

• Identify Business Needs and Solutions (26%) and Identify Data Needs (26%) together make up more than half the exam, so they warrant the largest share of study time — together over 50% of your prep hours.
• Operationalize AI Solution (17%) and Manage AI Model Development and Evaluation (16%) are mid-weighted and should each get a solid, but smaller, block of review.
• Support Responsible and Trustworthy AI Efforts (15%) is the lightest-weighted domain but still touches distinct, testable concepts (privacy, bias, compliance, documentation) that shouldn't be skipped just because its weight is lower than the others.

As a rough planning estimate, candidates with some existing AI/ML project background might budget somewhere in the range of 20–40 hours of focused review distributed across the five domains per the weights above, while those newer to AI project concepts may need more. Treat this as a starting point to adjust, not a benchmark.

How to prepare

• Start with the two 26% domains. Identify Business Needs and Solutions and Identify Data Needs are tied for the heaviest weight on the exam, so build your study plan around their task lists first — scoping, feasibility, ROI, risk assessment, and data sourcing/evaluation tasks (ECO-II and ECO-III) should be the material you revisit most often.
• Don't under-invest in the lightest domain. Support Responsible and Trustworthy AI Efforts carries the lowest weight (15%) but covers concepts — bias checks, transparency, audit trails, regulatory compliance — that can appear as standalone questions distinct from the technical domains. A quick read-through isn't the same as being able to apply ECO-I-T3 (bias checks) or ECO-I-T4 (regulatory and policy compliance) to a scenario.
• Study the project lifecycle in order. The five domains roughly trace the arc of an AI initiative: define responsible-AI guardrails, identify the business need, identify the data, develop and evaluate the model, then operationalize it. Reviewing the tasks in that sequence (ECO-I through ECO-V) can help the material click together rather than feel like five disconnected lists.
• Practice applying go/no-go judgment. Several tasks across Manage AI Model Development and Evaluation and Operationalize AI Solution (e.g., ECO-IV-T5, ECO-IV-T6, ECO-V-T3) are about making governance and readiness decisions, not just technical steps. Practice questions that frame these as scenario-based decisions tend to be more useful than flashcard-style recall.
• Use practice questions written from the public blueprint. CandiCerts practice questions are written independently from the public exam blueprint, not sourced from the certification body's proprietary exam content. Use them to check whether you can apply each domain's tasks to realistic scenarios, not just recite definitions.
• Confirm logistics directly with PMI. Pricing, registration windows, and eligibility requirements change over time and are not part of the content outline. Check PMI's official PMI-CPMAI page for current figures before you schedule your exam.