Over the past year, as I’ve collaborated more deeply with generative AI tools, I’ve come to a simple but powerful realisation:

Context is everything.

You can write the most technically precise prompt in the world, but if you don’t give the AI enough background, story, framing, or clarity about what really matters, it’s like briefing an eager graduate assistant with no life experience. They’ll jump to help – but they’ll often miss the point.

That’s why, for me, context engineering has overtaken prompt engineering as the more important skill for leaders and thinkers working with generative AI. And it’s why I now use two powerful models – CPQQRT and CRAFT – to frame my own thinking and help others get the best out of these tools.

The Limitations of Prompt Engineering Alone

Let’s be clear – prompt engineering matters. It’s how you interface with AI systems, and it often involves clever ways to extract better responses. Think of it like asking better questions in a coaching conversation.

But the early fascination with prompt engineering has led to some misunderstandings:

• • People chase “magic prompt templates” instead of learning how AI actually works.
  • They treat the tool like a vending machine – insert clever request, get shiny answer.
  • They forget that generative AI lacks lived experience, ethical reasoning, or intent.

Without context, AI becomes a high-speed content generator – great at volume, poor at value.

What Is Context Engineering?

Where prompt engineering focuses on what you say, context engineering is about what the system understands before you say it. It’s the discipline of embedding:

• • Background information
  • Objectives and constraints
  • Tone, format, and audience
  • Examples, stories, and history
  • Relevance, urgency, and nuance

In short, context engineering mirrors the way experienced leaders brief their teamsnot just with instructions, but with the full picture.

“If I had more time, I’d write a shorter prompt,” someone might say. But the truth is: If you give better context, you get better outcomes.

The CPQQRT Model: Briefing Like a Leader

When I coach leaders on task delegation, I teach a model called CPQQRT. I first came across it while working as Head of Organisational Development at SOHAR Aluminium in Oman. As part of Rio Tinto’s Management Operating System (MOS), this model was used to develop leadership and operational discipline – particularly in complex, high-capital environments.

CPQQRT stands for:

• • C – Context: Why the task matters. What’s the backstory?
  • P – Purpose: What is the objective? What does success look like?
  • Q – Quality: What are the minimum standards of acceptable/unacceptable?
  • Q – Quantity: How much is needed? What are the resource or cost expectations?
  • R – Resources: What’s available? Time, labour, money, knowledge, data, policies, procedures?
  • T – Timeframe: By when? What deadlines or delivery windows apply?

When you frame your AI prompts with this in mind, you move from a transactional prompt to a strategic brief. And the shift in the quality of the output is immediate.

The CRAFT Model: Structuring a Clear Prompt

While CPQQRT helps you think like a leader, another model – CRAFT – helps you shape your initial AI prompt.

Developed by Brian Albert, founder of Lawton (lawtonlearns.com) in Denver, Colorado, CRAFT is a practical framework designed to eliminate vague, unproductive prompts and replace them with clear, outcome-driven instructions.

CRAFT stands for:

• • C – Context: Provide relevant background or overview.
  • R – Role: Specify who the AI should act as (e.g. expert, writer, analyst).
  • A – Action: What do you want AI to do? Break it down clearly.
  • F – Format: Define the structure of the output (e.g. table, bullets, blog).
  • T – Target Audience: Who is the output for? What tone or depth is needed?

“AI isn’t useless. You’re just asking it weak questions.” – Brian Albert, Lawton

CRAFT isn’t about tricking the system – it’s about making your intent explicit. It’s a clear, professional way of briefing AI as you would a junior team member.

These Models Work Together – Not in Competition

Some people talk as if you have to choose between prompt engineering and context engineering. That’s a false choice.

• • Prompt engineering is the technique.
  • Context engineering is the thinking that informs it.

You need both – but context always leads. Especially when the goal is enduring, safe, human-centred outcomes. The AI can assist with logic, language, and speed. But it’s the human who must bring the judgment, ethics, and clarity of intent.

Think of CRAFT as your first draft. Use CPQQRT to refine it into something truly fit for purpose. 

Working with AI Is a Thinking Partnership

The best way I can describe working with AI is this:

It’s like briefing a smart, enthusiastic, sycophantic graduate executive assistant with no life experience.

The assistant is never going to challenge your assumptions. They’ll never push back or ask, “Are you sure that’s the best course of action?” They’ll just do what you say. If you give them half the story, you’ll get half the value.

That’s why context matters. That’s why CPQQRT matters. That’s why leaders need to build their skill not just in technology – but in how they think, brief, and iterate with intelligence systems.

Five Takeaways for Leaders Exploring AI Collaboration

1. 1. Frame your prompt like a leader would brief a team. Set the scene, define success, and specify constraints.
   2. Use CPQQRT to clarify your own thinking before you even open the AI tool.
   3. Write your first request using CRAFT – make it clear who the AI should be, what you need, how to format it, and who it’s for.
   4. Treat the AI as a collaborator, not a content machine. Think aloud, test ideas, build iteratively.
   5. Remember: you are responsible for the output. AI may generate the words – but you own every one.

Closing Thought: The Future Belongs to Contextual Thinkers

Prompt engineering opened the door. Context engineering invites the brilliance in.

As I prepare for my PhD on how we harness algorithmic systems to help leaders build thriving organisations, I’m convinced this is just the beginning. Models like CPQQRT and CRAFT aren’t just about briefing AI. They’re about equipping humans to lead in the age of intelligent systems – with clarity, ethics, and practical wisdom.

And that, in the end, is the work that matters most.