Director, Learning Design & Development| PMIAsheville, NC, United States
With Generative AI, iteratively refining and optimizing prompts can lead to better AI-generated results. This may involve adjusting the specificity or clarity of the prompt to increase relevance and accuracy of results.
What examples do you have of how improving a prompt drastically changed the output quality? What specific changes did you make that led to the improvement?
Cornel Almondproject manager| officialAppsSan Diego, USA
One example from my own work: I use AI to tailor my resume for project manager applications. My first prompt was "tailor my resume for this job posting" and the output was generic, with reworded bullets that added nothing. The change that fixed it: I added hard constraints and one example. I named the exact sections to keep, banned invented metrics, and pasted one bullet in the format I wanted. The output went from unusable to something I could submit after a quick review. The lesson I took: telling the model what to preserve matters as much as telling it what to produce. Saving Changes...
I had asked AI to generate a status report based on the most recent sprint. I failed to define the audience, indicating that names of assignees would not be appropriate for a senior level report. I also asked that it not focus on tasks but features and their progress. Just a few changes gave much better results. The initial prompts can be frustrating, but we are all learning together (and by "we" I mean the tool and ourselves)! Saving Changes...
In my experience, refining a prompt can transform a generic or superficial response into a structured, actionable outcome. The biggest improvement comes from adding clear context, defining the objective, specifying constraints, and establishing the expected output format. For example, asking AI to “analyze project risks” may produce generic risks, while providing the project context, risk criteria, and requesting evidence-based recommendations produces much more relevant results. I see prompt refinement as an iterative discipline: evaluate the output, identify gaps, adjust the prompt, and repeat until the result aligns with the project objective.
One problem I noticed when using AI was that it agreed with me too often. If my idea had a weak point, it didn't always challenge it. It could also give a fact and an assumption with the same level of confidence.. So I started giving the AI two simple instructions: challenge my thinking before answering, and tell me what I might be missing. I also ask it to separate what it knows from what is probable or just an assumption. This has worked well for me. I get more useful pushback and I know what I need to verify. Saving Changes...
I am coming across these different prompt engineering formulas for the first time. I will certainly use them in my market research project for a new product I'm exploring. Saving Changes...
There are framewoks to create prompt. This is part of the Prompt Desing discipline. Those that gave me and the initiatives where I was included are:R-T-F (Role-Task-Format), T-A-G (Task, action, goal), B-A-B (Before, after, bridge), C-A-R-E (context, action, result, example), R-I-S-E (role, input, steps, expectations).
Creating plugins with special tasks that I already went through using AI, I collect the whole prompt iterative steps & use the same as a plugin for future use. Saving Changes...
div class="ql-code-block-container" spellcheck="false"div class="ql-code-block" data-language="plain"My experience is quite limited; however, I believe that as PMs learn to develop clear and detailed prompts, the outcome will be beneficial for our role/div/div Saving Changes...
PMI Team Member
