Just as humans become more helpful when they’re given clear instructions by a skilled manager, modern AI tools produce more useful results when the person at the keyboard knows how to deliver directions deliberately.
For the latest installment of our new Driving Innovation series, we reconnected with Jordan Goure, recurrent entrepreneur and co-founder of the dynamic hiring platform Picsume. According to Goure, consumer-facing AI-powered chat apps like ChatGPT, Gemini, Claude, or Perplexity become dramatically more useful when their users learn a single critical skill: prompting.
The go-to verb tech industry experts use when describing AI interactions, prompt doesn’t always mean “right away.” To prompt, in plain language, is to cause or encourage an action. Imagine a little poke or prod or nudge in the right direction. While it feels natural to write your favourite AI chatbot a question, even a polite request implies the possibility of a negative answer. Remember: You’re in charge, here. You’re not asking the tool what to do; you’re telling it.
Recipe for Results: Jordan Goure’s Simple Prompt Framework
You don’t need to reinvent the wheel each time you want to give the AI machine a little spin. The following basic fill-in-the-blanks template (or “structure,” in Goure’s words) makes a good starting point for most prompts:
Act as a ___. Help me with ___. The goal is ___. Here’s the context: ___. Please format it as ___.
Suddenly, harnessing the power of AI feels practical and repeatable.
While generative AI tools can stumble into effective results when given broad commands or queries, they do their best work when instructed to roleplay within a clear, well-established framework. I like to think of something like ChatGPT as a limitless staff of well-meaning and industrious but occasionally dopey personal assistants. It’s easy to imagine use cases for an army of aides like that, but a thoughtful leader wouldn’t leave them up to their own devices.
According to Goure, AI tools respond well to direction, perspective, structure, and clarity. When you tell an AI how to act or “think,” you take steps toward shaping the tone, structure, and depth of its response. “Instead of asking vague questions like, what should I say?” he explains, “guide the AI by assigning it a role or establishing a clear framework.”
For example, before your request a response from your AI tool, consider adding an additional command:
- Think like a conflict resolution specialist.
- Respond using principles from nonviolent communication.
- Approach this using cognitive behavioural therapy techniques.
- Act as an executive coach helping me prepare for this conversation.
“By assigning a role, a psychological framework, and a specific lens,” adds Goure, “you can influence and optimize tone, structure, depth, and practicality.”
Quick hits: Four Pro-Tips for Productive Prompting
- Provide clear and coherent context.
Who is involved? What is at stake? Should the AI “know” any relevant history before it generates a response?
- Order extra helpings.
Instruct your AI tool to produce multiple options, each with a slightly different quality. Trust yourself to choose the best one. You’ll feel like a CEO assigning three direct reports to the same project. For example:
Give me three approaches: direct, empathetic, and firm.
- If at first you don’t succeed, request revisions.
Engineers understand the difference between a prototype and a market-ready product; writers appreciate the chasm between a first draft and a publishable, polished piece. Even if you’re broadly happy with your initial prompt’s results, some straightforward revision instructions will likely lead to an even better, bespoke-seeming outcome. Consider:
Make this calmer.
Shorten this to four sentences.
Make this sound more confident.
- Iterate intentionally.
Remember: Rome wasn’t built in a day. While AI can significantly speed up scores of common tasks, it doesn’t promise its best possible results on the first try. If you expect the end destination to require a few detours, you won’t feel like you’re wasting time on the journey.
According to Goure, we’ll all do a bit better if we move away from thinking about AI-generated product in binary (i.e., correct and incorrect) terms. “Prompting isn’t about getting the ‘right’ answer immediately,” he cautions. “It’s about collaborating with AI as a structured thinking partner. When you provide clarity, constraints, and perspective, the quality of output increases dramatically.”
