Sponsored by
AI Needs Experience
Companies are betting on AI to replace deep product knowledge.
I think that's one of the most expensive mistakes a product leader can make.
Not because AI isn't powerful. It is. But because there is a ceiling to what it can do — and that ceiling is built from something AI cannot access: years of being inside the product.
A new study from Stanford and Harvard1 proved it. They ran a controlled experiment at a real fintech company. Their finding? AI narrows the skill gap between novices and experts. It does not close it. They named this limit the AI wall.
I've seen this wall up close. And the day I understood it, someone had already walked out the door.
In today's newsletter:
What the Stanford & Harvard research actually found
A real story from a B2B design team, and the designer who made AI powerful
Why your rock stars are your most irreplaceable knowledge asset
Three things you can do before the next resignation
AI Amplifies What You Know. It Cannot Create What You Don't.
Here is what the research found.
Researchers divided employees into three groups: expert writers, marketing specialists with no writing experience, and technologists with no marketing background at all. Some received AI tools. Some didn't. IG executives then rated the results.
Without AI; experts outperformed everyone. Expected.
With AI; the gap narrowed, especially for the marketing specialists. They almost matched the expert writers.
But the technologists? Their scores barely moved. With or without AI.
Why?
Because they couldn't judge the AI's output. They didn't know what to keep, what to cut, what sounded off. They copy-pasted. They had the tool but not the taste.
The researchers concluded: the further you are from the domain knowledge, the less AI can help you.
AI amplifies what you already know. It cannot create what you don't.
Now think about this in the context of your product team.
A junior designer using AI can produce faster mockups. More iterations. Better-looking output.
But can they answer: why did we design it this way? Why was that feature cut? What did the client actually mean when they pushed back on that workflow?
No. Because AI was not in those meetings.
She Was Great With AI. That's Exactly Why Her Departure Hurt.
I want to tell you about a designer I worked with.
She joined a B2B company as one of the very first members of the design team. Before there was a design system or formal processes. She was there when the product was still finding its shape.
Five years later, she left. As, she was simply ready for a new chapter.
What made her extraordinary wasn't just her skill. It was her memory.
She knew every major product decision — and the reasoning behind each one.
She knew
Which features had been debated for months before being cut.
Which client relationship had shaped an entire workflow.
The why behind the what.
And here's the part that matters most for this conversation: she was great with AI.
She wasn't resistant to it. She embraced it. Because she had five years of product knowledge, she could direct AI with precision.
She knew when the output was wrong. She could push it further, refine it faster, and catch what it missed.
She was exactly the kind of expert the HBR researchers described — someone whose deep domain knowledge made AI more powerful, not less.
When she left, the team kept the same AI tools. Same access. Same subscriptions.
But suddenly, nobody could answer the most basic question a designer needs to answer:
Why did we build it this way?
That answer had a name. And she was gone.
The AI wall hit immediately.
Not because the tools got worse. Because the knowledge did.
What Kim Scott Gets Right
In Radical Candor, Kim Scott describes two types of high performers.
Superstars are on a steep growth trajectory. They want more responsibility, new challenges, promotions.
Rock stars are on a gradual growth trajectory. Not climbing. Not restless. Just deeply, quietly excellent at what they do.
Rock stars are your institutional memory.
They hold the full context of why your product is what it is. They don't need a new title to stay engaged.
They need ownership, depth, and the feeling that their expertise is valued.
The designer I described was a rock star.
She wasn't trying to become head of design. She just knew the product better than anyone. And because she wasn't asking for a promotion, it was easy, dangerously easy — to assume she was stable. That she would always be there.
And here is where AI makes this problem worse.
When leaders believe AI can close the knowledge gap, they feel safer letting rock stars leave. They think:
we have the tools, we have the documentation, we'll be fine.
The research says otherwise. The AI wall is not just about task performance. It is about context. And context takes years to build.
Your junior team members, no matter how good their prompts are, cannot recover what a five-year rock star carries in their head.
How Can We Solve This
Three things I believe every product and design leader should do.
1. Identify your rock stars — before they plan to leave.
Look around your team. Who has been there the longest? Who knows the why behind every major decision? Who never asks for a promotion but is somehow always the person others go to when they need to understand the product?
That person is a rock star. And they may not be loud about what they need.
Check in. Recognize them explicitly. Reward them with autonomy, ownership, and depth — not just titles. Don't wait until their resignation to realize what they were holding.
2. Make knowledge visible — not just written.
A Notion or Confluence page tells you what was decided. It rarely tells you why.
Record a Loom. Walk through a feature and explain the problem it was solving. No script. Just your voice and your screen. Encourage your team members to do the same.
Record your retrospectives. That is where the why lives — in the conversation, not the summary.
Five minutes of recording can save weeks of confusion.
3. Stop expecting AI to compensate for attrition.
AI is a tool for speed. It amplifies existing knowledge. It does not generate institutional memory from scratch.
The HBR researchers put it simply: "Expertise is irreplicable. No technology can substitute for it."
I believe that. And I'd add:
The people who carry that expertise are often the quietest ones in the room. The ones who've been there the longest. The ones who don't need a new title to show up fully.
Key Takeaway
Before you build your AI strategy — look around. Who on your team holds the memory of your product?
Make sure they know you see them.
AI Tools I'm Using This Week
🤖 Claude — Writing articles and documentation has never been this fast — the thinking stays mine, and Claude Cowork handles the rest.*
🎙️ Wispr Flow — I speak four times faster than I type. This article? Wispr Flow had a big contribution. If you are still typing everything, you are leaving a lot of time on the table.*
🎨 Gamma — Yesterday I built a full presentation from scratch in two hours. The graphs, the layouts, the creativity you can develop in there — it is unmatched. What used to take a full day now takes one focused session.*
* These software are not sponsored
You don't need to be technical. Just informed.
Most AI newsletters are written for engineers. This one isn't.
The AI Report is read by 400,000+ executives, operators, and business leaders who want to know what's happening in AI — without wading through code, jargon, or hype.
Every weekday, we break down the AI stories that matter to your business: what's being deployed, what's actually working, and what it means for your team.
Free. 5 minutes. Straight to the point.
Join 400,000+ business leaders staying ahead of AI — without the technical overwhelm.