10 golden rules and four principles for AI prompting

Want to get more out of artificial intelligence (AI)? Knowing how to use these tools correctly will give you an advantage.

AI prompting is emerging as a core capability for project delivery professionals – not in a technical sense, but as a way of structuring thinking. It is about being clear on intent, giving the AI tool enough context and shaping the response so that it is usable in the environment we work in.

In many ways, this mirrors what we already do. We define scope, clarify requirements, provide context and iterate until something is fit for purpose. The same disciplines apply here. The professionals who get the most from AI are those who bring the same rigour to a prompt that they would to a project brief.

Once you start using these tools more regularly, a few principles become clear:

Be clear on what you are trying to achieve. If the task is not well defined, the output rarely is. A prompt such as “summarise this project” tends to result in something generic. When you shape it more deliberately – for example, asking for a summary for a sponsor that focuses on risks and decisions in a concise format – the response becomes far more usable.
Provide context. The model does not know your project, organisation or constraints unless you explain them. Even a short line about the type of project, the stage you are in or the key challenge you are facing can change the quality of the response.
Define the role. Telling the model who to act as shapes the perspective of the response. Asking it to respond as a project sponsor will produce a very different output than asking it to respond as a contractor or a PMO lead.
Specify the output format. Asking for bullet points, tables, decision logs or executive summaries makes it much easier to use the response in a project setting. Structure improves usability.

There are several frameworks that help apply these principles consistently. CRIT (Context, Role, Instruction, Tone), RACE (Role, Action, Context, Expect) and GCSE (Goal, Context, Source, Expectation) all offer slightly different structures, but the underlying idea is the same. A simple prompt template that works well in practice is: “Act as [role]. Context: [project situation]. Task: [what you need]. Output: [format].”

Use AI as a thinking partner

One of the most valuable approaches is to treat AI as something you partner with rather than something you simply use. The value comes from using it to explore options, test ideas and challenge thinking, rather than expecting a single answer. In practice, this often looks like a conversation: you start with a prompt, get a response, then refine it.

For example, instead of asking “What are the key risks on this project?”, you might say: “Identify the key risks on this project, then challenge your own analysis by highlighting assumptions, gaps and areas of uncertainty.” This shifts the interaction from generating an answer to improving the quality of thinking.
Start the right way on your AI prompting journey, then follow these 10 golden rules to continue to get the best out of these intelligent tools:

The 10 golden rules for AI prompting

Be clear on intent. If you are not clear, the output will reflect that.
Provide context. The model does not know your project unless you tell it.
Define the audience. A sponsor update is very different from a team update.
Structure your prompt. Role, task and format go a long way.
Iterate and refine. One-shot prompts rarely give a considered answer.
Challenge outputs. Do not accept something just because it sounds plausible.
Break down complexity. Smaller prompts lead to better results.
Repeat what matters. Restate key constraints in longer conversations.
Use AI within your competence. You must be able to judge the output.
Verify before use. Always check before relying on outputs in a real project context.

Keep this expert advice close at hand, and the next time you turn to AI for help, you’ll achieve better outcomes from your prompting. Good luck!

APM members can get exclusive access to the new APM Learning module ‘Prompt engineering for project professionals’