How we prompt, collaborate, and think with machines—starts here.
When people talk about “prompt engineering,” most of the time they’re missing the point.
They treat it like an art of copy-paste hacks, fill-in-the-blank templates, and tricks to get better responses from ChatGPT. And while those tips have their place, they’re firmly stuck at what I call Level One—and they’re missing the real power of this moment.
Prompting isn’t just about talking to AI. It’s about learning to think clearly, structure ideas, and engage in meaningful creative dialogue. And when you learn to move from Level One to Level Two, that’s when everything changes.
What Is Level One Prompting?
Level One prompting is all about clarity and context.
At this level, you learn how to:
•Set a clear role for the AI
•Define the task or goal
•Add relevant context
•Specify the desired output format
(That’s what the COAT framework helps with—if you haven’t seen that, check out the COAT video or blog post.)
Level One is essential. You can’t collaborate with an AI if you’re not expressing yourself clearly. It’s like walking into a room with your team and not knowing what the meeting is about.
But here’s the thing: Level Two isn’t just better prompts. It’s a different mindset.
What Is Level Two?
Level Two is where you stop treating AI like a tool—and start treating it like a colleague.
You stop writing prompts like commands, and start engaging in conversations. You stop obsessing over perfect wording, and start co-creating, iterating, reflecting.
Here’s a story from my own work.
I was building a Prompt Manager app using Cursor, and Claude Sonnet 4 was the paired model behind the scenes. I gave it a simple instruction: “Let’s build this test-first.” Claude replied, “That’s actually my preferred way to work.”
That’s when I knew I wasn’t just using an LLM—I was collaborating with a design partner.
We didn’t just generate code. We discussed architecture. We refactored early and often. We extracted business logic from the UI and made it testable. Claude recognized patterns. It anticipated the need for modularity. It co-designed.
I’ve written code for decades. But that session felt like a breakthrough.
This is what Level Two feels like.
The Mindset Shift: From Control to Collaboration
At Level One, the focus is on getting what you want.
At Level Two, the focus becomes discovering what’s possible together.
It’s a shift from output to insight. From scripting answers to surfacing assumptions. From “do this for me” to “let’s figure this out.”
That shift doesn’t just happen in language. It happens in the brain.
Why Anthropomorphism Works
When we treat AI like a tool, we stay in a narrow cognitive lane—focused on syntax, command structures, and expected output. But when we treat AI as a collaborator—even a fictional one—we activate different parts of the brain.
Neuroscience suggests that different brain regions are engaged when we interact with perceived agents versus inert tools. The medial prefrontal cortex, for example, lights up during social cognition—when we’re predicting intent, modeling others’ thinking, or co-navigating decisions.
So yes, it might feel a little strange to say, “Claude prefers TDD.” But that anthropomorphism unlocks creativity. It raises our expectations. And surprisingly often, the AI rises to meet them.
Level Three and Beyond (A Glimpse)
We’re just beginning to explore Level Two. But there’s more ahead.
Level Three is about pattern-based co-creation. It’s where we teach the AI our mental models and domain rules and then co-develop solutions that reflect that shared understanding. It’s where collaboration becomes orchestration. And that’s a topic we’ll dive into soon.
For now, the most important thing is this:
The future of work isn’t in prompting. It’s in partnering.
How to Start Leveling Up
To move from Level One to Level Two:
1. Master the basics. Learn to write clear prompts using COAT. Make your intent and context explicit.
2. Think out loud. Ask AI to reason with you, not just respond to you. Let it reflect, reframe, and challenge.
3. Expect more. Treat the AI like a thoughtful peer, and you may be surprised what comes back.
Final Thoughts
Prompting well is a valuable skill. But prompting with presence, with curiosity, with a willingness to be in dialogue—that’s where the magic is.
If you’ve only ever copied prompts or scripted your way through AI tools, you’re missing the point.
You’re not just using AI.
You’re teaching yourself how to think more clearly, design more cleanly, and collaborate more creatively.
So… what level are you playing at?
