Author: Daria Frey, PMP
I won’t spend time telling you how AI is transforming project management, you already know that. But here’s something worth thinking about: the way you ask AI for help can completely change the results you get. Let’s look at an example.
Two project managers, Kai and Gerda, need a project risk analysis for an upcoming initiative. They both use ChatGPT, but their approaches differ.
Kai’s prompt:
"Provide a risk analysis for Phase 1 clinical trial supply project".
Gerda’s prompt:
"Act as a senior risk management consultant with 20 years of experience in clinical trial supply. Generate a structured risk analysis for a clinical trial supply process for a Phase 1 study conducted in the US, with supply manufacturing in Canada. Identify at least ten key risks across logistics, regulatory compliance, quality control, and operational efficiency. For each risk, classify its severity (low, medium, high), describe its potential impact, suggest specific mitigation strategies (e.g., contingency planning, supplier audits, cold-chain validation), and align recommendations with GMP, GDP, ICH Q9, and applicable FDA and Health Canada regulations. Format the analysis as a structured report suitable for clinical trial managers and regulatory compliance teams, using clear, professional language. Where applicable, reference dependencies on CDMOs, third-party logistics providers, and clinical trial sites. Provide the risk analysis in a table format with columns for “Risk Category”, “Severity Level”, “Potential Impact”, “Mitigation Strategy”, and “Industry Best Practices".
You can imagine how different the responses would be. Try it yourself and see.
This example highlights a simple but crucial takeaway: The quality of AI output is directly dependent on the quality of the input. This is the essence of prompt engineering – the ability to craft precise, strategic instructions that guide AI systems to deliver meaningful insights and actions.
I hope this demonstrates that to use AI at its full capacity, you need to master your prompts. Now, let’s break down the core principles.
Crafting Effective Prompts: Core Principles
- Defined Role or Perspective – Assign a role to AI for more contextually accurate responses.
Example: “Act as a project risk analyst and identify potential cost overruns” - Clarity & Specificity – Clearly articulate the task, avoiding vague or overly broad requests.
Example: “Generate a project timeline for a new product launch, including key milestones and dependencies” - Contextual Relevance – Incorporate industry-specific guidelines, frameworks, or past project data.
Example: “Analyze risks based on the PMI risk management framework” - Tone and Audience Consideration – Specify the expected tone, depth, and target audience.
Example: “Summarize this project update for senior executives in a concise, data-driven format” - Use of Examples – Provide examples to clarify expectations and guide AI interpretation.
Example: “Provide a Gantt chart template similar to [insert known example]” - Ethical Considerations & Compliance – Ensure compliance with company policies and ethical AI use.
Example: “Generate a stakeholder communication plan while adhering to GDPR data privacy guidelines”
You can check Gerda’s prompt above as a good example that utilises these principes.
Furthermore, to master your prompt engineering skill, avoid these common mistakes:
- Information Overload – Keep prompts focused, avoid asking for too much at once;
- Vague Requests – Generic prompts lead to generic (and often unhelpful) responses;
- Skipping Validation – AI-generated outputs should always be reviewed against project goals and industry standards.
Remember, AI-generated responses improve with refinement. After receiving an initial output, refine your prompt using:
"Review your response considering: accuracy and alignment with project goals, risk assessment completeness, and resource optimization opportunities"
Additionally, to allow AI to probe for missing details, ensuring a more tailored and useful output, include this phrase:
"Ask me clarifying questions to provide the most effective response"
AI can make mistakes, therefore always follow the V3 Principle:
- Verify against current industry standards
- Validate with subject matter experts
- Version Control successful prompts for future use
The second most important rule? Data security is non-negotiable – never share sensitive or proprietary information with AI systems.
If you’re interested in diving deeper into prompt engineering, I highly recommend the PMI course “Talking to AI: Prompt Engineering for Project Managers”.
Ultimately, the level of detail in your prompt should match the complexity of the task. If you're simply looking for basic information, a short prompt may be enough. But when tackling high-impact tasks, it's worth leveraging AI’s full potential with a well-structured prompt.
As AI’s role in project management continues to expand, we, as committed project management professionals, must stay ahead of the curve by using AI to:
- Faster, smarter decision-making
- Greater efficiency in daily tasks
- More strategic and impactful leadership