After working with a few different teams on AI adoption, I keep seeing the same patterns.
Companies roll out AI tools. Output volume increases. But the substance quietly erodes —
the kind of critical thinking only employees carry because they understand how the organisation actually operates.
When people get AI tools, they unknowingly start replacing their thought process with AI
rather than using AI to execute their thought process. These are obviously two very different things —
but not always clear when everyone has deadlines to meet.
"AI replacing your thinking vs. AI executing your thinking — two very different things."
The Pattern — Which one is your team?
📋
Task arrives
🤖
Open AI immediately
📋
Copy output directly
📤
Submit
⚠️
Volume increases. But output loses the nuance of how your organisation actually operates. Critical thinking atrophies. AI is replacing, not executing, their thought process.
🧠
Think first — no AI
🤖
AI executes + expands
✍️
Add fingerprint layer
🏆
Submit
✅
Output is fast and retains organisational intelligence. Your team's thinking grows sharper — AI accelerates it, not replaces it.
To fix this, I've tried multiple approaches. One that has consistently worked is asking teams to follow a 3-step rule I call the AI Sandwich.
The 3 Layers — Click to explore each one
The AI Sandwich Framework
🧠
Humans Think First
Before opening any AI tool, map out the problem in your own words. The thought process stays with you. Writing better prompts is also part of this layer — a good prompt is already half the thinking done.
This layer is non-negotiable. Skipping it is where most teams go wrong — they open ChatGPT before they've even defined what problem they're solving.
🤖
AI Does the Heavy Lifting
The filling isn't a single step — it's a loop. Once you bring AI in, you cycle through three moves until the output is solid enough to fingerprint.
1
Craft Better Prompts
Your Layer 1 thinking feeds this. A well-framed prompt is context + intent + constraint — not just a question. The better the prompt, the less looping you need.
2
Brainstorm to Uncover Blind Spots
Ask AI to challenge your framing, suggest what you missed, or argue the opposite. This is where AI earns its value — not drafting, but surfacing what you didn't think of.
3
Adjust Scope — In or Out
Based on what emerged, decide what to add and what to cut. Scope creep kills quality; scope clarity sharpens it. This step loops back to step 1 until the output is ready to fingerprint.
Loop until output is ready for Layer 3
The AI doesn't own the thought. It's executing your thought — at scale and speed. That's the key mental shift teams need to make.
✍️
The Fingerprinting Layer
This is where you make the final edits. Go back and add what AI couldn't — the context, the culture, the signature of how your organisation works. This layer is what makes the output less robotic and more trustworthy internally.
This layer makes your output a genuine organisational asset — not just another AI response. It's the difference between a document people trust and one they immediately distrust.
It works because it is simple and gives the team a mindmap that is applicable across functions.
This way they never fully hand over their thinking to a machine.
Quality vs. Volume — Simulate It
Output comparison — with vs. without the sandwich
Without Sandwich
Output Volume
Critical Thinking
Org Context Retained
Output Quality
With AI Sandwich
Output Volume
Critical Thinking
Org Context Retained
Output Quality
Volume stays comparable — the sandwich doesn't slow you down. The gains are in quality, retention of thinking, and organisational trust in the output.
Build Your Org's Fingerprint Layer
What does your org's Layer 3 look like?
Select the fingerprints that apply to your team to see how they shape Layer 3…
Try this out if you have rolled out AI to your teams. You will be amazed at the results.