By Markus Kopko, PgMP®, PMP®, PMI-CPMAI™
March 26, 2026
Most project managers who try AI for the first time ask something like, “Create a risk register for my project.” The AI responds with a generic list of risks that could apply to any project in any industry. The project manager concludes that AI is not useful for real PM work. The problem was never the AI. The problem was the prompt.
Prompting is the interface between what you know about your project and what the AI model can do with that knowledge. A vague prompt produces a vague output. A structured prompt produces output that is specific, relevant, and worth reviewing. The difference is not talent. It is technique.
Why Generic Prompts Fail in Project Management
AI models respond to the information they receive. They do not know your project, your stakeholders, your constraints, or your delivery history. When you write “Draft a status report for my project”, the model fills the gaps with assumptions that are statistically probable but contextually wrong.
Project management is a context-dependent discipline. Every risk assessment depends on the specific project environment. Every stakeholder communication depends on relationship dynamics.
Every schedule decision depends on resource availability and organizational priorities that exist nowhere in the model’s training data. The project manager is the only source of that context. Prompting is the mechanism for transferring it.
The CRISP Framework for Project Management Prompts
| The CRISP Framework for Project Management Prompts | |
| Context | Provides project-specific information: industry, phase, team size, constraints |
| Role | Defines the perspective: risk analyst, communications specialist, scheduling expert |
| Instruction | States what you need: draft, analyze, compare, identify, prioritize. |
| Scope | Sets boundaries: which deliverables, time period, stakeholder group |
| Parameters | Define format (table, bullet list, paragraph), length, level of detail, and tone. This is where you control how the output is structured, not just what it covers. |
A prompt without Context produces generic content. Without Role, unfocused content. Without Scope, content that tries to cover everything and covers nothing well. Each element serves a specific function.
Applying CRISP to Real Project Management Tasks
Let’s cover two examples that demonstrate the difference between an unstructured prompt and a CRISP-structured prompt for common PM tasks.
Example 1: Risk Identification
Unstructured: “List the risks for my IT project.”
CRISP: “Context: We are in month 3 of a 12-month ERP implementation for a mid-size manufacturing company. The project team consists of 8 internal staff and 4 vendor consultants. We have completed requirements gathering and are entering the design phase. Two key SMEs are shared with another project at 50% allocation. Role: Act as an experienced risk analyst for enterprise IT projects. Instruction: Identify the top 10 risks for the next project phase, including both technical and organizational risks. Scope: Focus on the design and build phases only. Exclude go-live and post-implementation risks. Parameters: Present each risk with a brief description, likely impact on schedule or budget, and one suggested mitigation action. Use a table format.”
The unstructured prompt above produces a textbook list. The CRISP produces a risk assessment that reflects your actual situation, the resource constraint you manage, and the phase you are entering. The output still requires review, but the starting point is substantively closer to usable.
Example 2: Stakeholder Communication
Unstructured: “Write a project status email.”
CRISP: “Context: Our data migration project is two weeks behind schedule due to unexpected data quality issues in the legacy system. The executive sponsor has expressed concern about the delay at the last steering committee meeting. We have identified a recovery plan that adds two resources for four weeks. Role: Act as a senior project manager communicating to an executive audience. Instruction: Draft a concise status update email that acknowledges the delay, explains the root cause without being defensive, presents the recovery plan, and requests approval for the additional resources. Scope: Address the schedule impact and recovery only. Do not cover budget, scope, or other project dimensions. Parameters: Keep the email under 250 words. Professional tone, direct language, no jargon.”
The unstructured prompt produces a template with placeholders. The CRISP prompt produces a draft that addresses your situation, your stakeholder’s concern, and the decision you need. You still own the final message, but you saved 20 minutes of drafting.
Three Principles That Make Prompts Work
Beyond the CRISP structure, three principles consistently improve output quality.
First, treat every prompt as a briefing. Be specific. Provide the relevant facts. State what you need. If you would not walk into a meeting and say, “Tell me about the project,” do not write a prompt that says the same thing.
Second, iterate instead of starting over. AI conversations are cumulative. “Make the risk descriptions more specific to our manufacturing context” produces better results than rewriting the entire prompt. Iteration is faster and more targeted.
Third, never submit AI output without human review. AI-generated content is a first draft. It accelerates creation, but the project manager’s judgment determines whether the content is accurate, complete, and appropriate for the audience.
What Prompting Cannot Do
Good prompting does not turn AI into a replacement for project management expertise. Rather, good prompts turns AI into a more effective tool for project managers who already have expertise.
AI cannot assess whether a stakeholder’s concern is genuine or political. It cannot judge whether a recovery plan is realistic given the team dynamics you observe but cannot quantify. Research confirms that frequent AI use correlates with reduced critical thinking (Gerlich, 2025), the precise capacity project managers need most. Critical thinking are the judgment calls that require experience, emotional intelligence, and situational awareness—human capabilities that are needed more than ever.
The PMBOK® Guide—Eighth Edition captures this purpose directly. By focusing on value as a core outcome, AI-generated output is only worth producing if it supports better project decisions. A well-structured prompt delivers value by giving the model enough context to produce relevant, reviewable output. In contrast, a generic prompt wastes the project manager’s time with content that requires more editing than writing from scratch.
As a member of the Core Development Team for the PMI Standard on AI in Project, Program, and Portfolio Management, I see this principle reflected throughout the standard’s development. AI amplifies the project manager’s effectiveness, but the quality of that amplification depends directly on the quality of the input. Prompting is where that quality starts.
Start With One Task
The most effective way to build prompting skills is to start with one task you do repeatedly, such as a weekly status report, a risk review preparation, or a meeting summary. Apply the CRISP framework. Evaluate the output. Iterate on the prompt. Once you have a prompt that consistently produces useful output, save it as a template and move to the next task.
Within a few weeks, you will have a personal library of PM-specific prompts that save meaningful time on recurring work. That time can be redirected to the strategic and relational work where human judgment is irreplaceable.
Key Takeaways
- Generic prompts produce generic outputs. The AI model performs to the level of context you provide.
- Use the CRISP framework: Context, Role, Instruction, Scope, Parameters. Each element serves a specific function.
- Treat prompts like project briefings. The skills that make you effective in meetings apply directly to prompting.
- Iterate on outputs instead of rewriting from scratch. AI conversations are cumulative.
- AI-generated content is always a first draft. Professional review is a requirement, not an option.
- The PMBOK®Guide, Eighth Edition, focuses on value as a core outcome. A structured prompt delivers value. A generic prompt wastes time.
- Start with one recurring task. Build a personal prompt library over time.
This article is part of a series leading up to the IIL webcast “5 Steps to Integrate AI into Your PPM Practices: A Tactical Blueprint” on June 24, 2026. Register at https://www.iil.com/your-ai-advantage-practice-habit-strategy/
Markus Kopko, PgMP
Markus Kopko is a strategic project and AI transformation expert with over 25 years of experience in project, program, and portfolio management. He is a member of the Core Development Team for the PMI Standard on AI in Project, Program, and Portfolio Management and served on the PMI Review Team for the PMBOK® Guide, 7th Edition. Markus holds PMP®, PgMP®, and PMI-CPMAI™ certifications. He is a trainer and content creator for IIL. He delivers the course, Generative AI for Project Management, on IIL’s learning platform.
PMP®, PgMP®, and PMBOK® are registered marks of the Project Management Institute, Inc. PMI-CPMAI™ is a trademark of the Project Management Institute, Inc.
