How To Product-think When AI Builds At Lightning Speed
The One Feature AI Can’t Replace
Hey, I’m Karo 👋,
AI product manager and creator of StackShelf.app.
I build, write about, and overthink products for a living.
If you’re new here - welcome! Here’s what you might have missed:
Claude Skills Are Taking the AI Community by Storm
The Indie Builder Economy on Substack
Ask ten people what product thinking means, and you’ll get twelve definitions.
Beyond the books, infographics, and frameworks, the best insight comes from founders who built things that survived contact with users. From watching how they product-think in motion:
how they question their own logic,
how they trade the cool for the useful,
how they drop the ego to listen,
how they stay deliberate when everything screams ship faster.
Today’s post comes from Orel, software engineer, founder of WriteStack and a living proof that product thinking gets real only when you start building.
I’ve featured Orel before, and I keep following his work because he shares it with curiosity, honesty and the kind of humility that makes you pay attention. There’s a lot we can learn from his journey. Enjoy!
As I build WriteStack, I start with drafts vibecoded in Cursor, then I revise, polish and ship them.
One day I thought to myself: What if WriteStack had a ChatGPT inside, wired to data and context? That would be really cool.
I spent 20 minutes building a high quality prompt and 10 minutes later I had a chat feature working out of the box.
That was freaking crazy.
That was the high point of my enthusiasm for Cursor and for AI-assisted coding in general. What once took me days now takes just minutes, even for complex features.
But beneath the speed and ease of building with AI, there’s a side few people acknowledge.
The part that is much harder to build.
Your product thinking.
Product Thinking, Plain and Simple
Product thinking = intentional product decision-making.
It’s making a deliberate, informed next move and evaluating not just if something should be built, but what it changes or disrupts within the overall system.
How It Works in Practice
When you consider a new feature, ask yourself:
Why is it necessary?
What will it take?
How will it affect everything else I’ve already built?
Example: Say that I want to add a Freemium Tier to WriteStack.
At first glance, it feels simple: one quick prompt to Cursor, a few lines of code, and I’d have a working version in minutes.
But if I actually stop to think, things get complicated fast:
What limits should I set for free users?
How will this affect existing paying subscribers?
What happens to my conversion strategy?
Will it spike database load?
Will it increase costs?
How do I announce it without confusing current users?
Do I need separate onboarding and email flows?
That’s where product thinking kicks in.
Pausing to consider the ripple effects and making a deliberate call on whether the feature is worth building at all.
AI won’t do that for you. It just gives you the illusion of progress.
And that’s where many builders lose the game.
They stop feeling the cost of their decisions, because AI absorbs all the friction.
But that friction - the part that slows you down, frustrates you, forces trade-offs - is exactly what makes you good.
AI Erodes Product Judgement
We like to think AI makes us faster. But in many cases, it makes us slower and duller:
A 2025 METR experiment found that professional developers using AI coding assistants actually took 19% longer to complete familiar tasks.
A 2024 study on cognitive offloading found a strong negative correlation (r ≈ -0.68) between frequent AI use and critical thinking skills.
And a survey across Chinese universities found that nearly 69% of AI users reported increased laziness and weaker decision-making ability.
In plain terms: the more we rely on AI, the less our brain practices reasoning on its own.
While AI handles the coding, and gives us the illusion of progress, we stop reasoning about the why. We just react to the what.
The code runs. The UI renders. The output looks clean.
But you don’t really know why it works. Or whether it should.
Three Ways to Sharpen Product Thinking
For me, the only real antidote to bad product thinking is friction.
Every time I rush, I lose judgment.
Every time I slow down, I get it back.
Here’s how you can do it too:
1. Don’t Chase Features, Chase Impact
Before adding anything, I ask myself: Will this change the experience for the user in any meaningful way?
If the answer is yes, I build it.
If the answer is I don’t know, I design a way to learn.
If the answer is no, I drop it, even if it’s really, really cool.
2. Think in Systems
Every new feature you build touches something else in your product:
Add a new AI feature ➜ your compute costs spike
Add a free plan ➜ your onboarding flow changes
Add a referral program ➜ your support inbox will get new edge cases
Nothing exists in isolation, it all echoes somewhere else. Pausing and considering the effects is product thinking in action.
You don’t need an all-day session. Fifteen minutes of complete focus is plenty.
Try it. You’ll be amazed at the amount of data you get out of it.
3. Trust Your Gut
When you delegate everything to AI, everything starts to feel possible, and suddenly, it all feels necessary.
I know that feeling.
Try to resist it and let your own mind - and your gut - take over the work and decide what actually belongs.
This is where judgment lives: in the pause between can and should.
Practice that pause. It’s the only thing AI can’t do for you.
Conclusion
AI makes coding accessible to anyone. That’s incredible.
But here’s the trade: the easier it gets to build, the harder it gets to think.
Product thinking isn’t some abstract skill you learn from frameworks. It’s what happens when you slow down enough to question your own decisions, when you feel the weight of what you’re adding, when you remember that every feature costs something.
So keep the friction, it’s the feature that keeps your judgment sharp.
💡Product thinking is only as strong as your ability to test it - and that’s exactly where the right tools make all the difference. Today’s post is brought to you by Reforge Build, an AI prototyping tool for product teams. Prototype from your existing product with AI that understands your customers, features, and strategy.
Additional Resources
You Might Also Enjoy
Rediscovering Product Craft in the Age of AI by The Product Venn
Building AI in an Agentic AI World by Isabel Gan
Bad Product Model: No Plan, No Rhythm, No Excuses by Wafaey.
Productness and Lack Thereof by Juha Korpela
How Proper PMs Think About AI Product Strategy by Jason Knight
The Translation Layer Every Strong Product Strategy Needs by Gabrielle Bufrem
Vibecoding x Cybersecurity: Survival Guide by Farida Khalaf, skelly & Karo
The AI Upskill Playbook for Product Managers by The Product Journey
Community Updates
👉 Final 2 days to take part in the Attitude Pulse!
👉 Anfernee is working on AI Agents for Solopreneurs. What should they tackle next?
🚀 New launches on StackShelf:
Join hundreds of Premium Members and unlock everything you need to build with AI. From prompt packs and code blocks to learning paths, discounts and the community that makes it so special.
WHY SUBSCRIBE ・LEARNING PATHS・ PREMIUM RESOURCES・ TOOLS ・TESTIMONIALS ・LIFETIME MEMBERSHIPS ・TESTIMONIALS
 | A guest post by
|
Orel, your three ways to sharpen product thinking are amazing! You and Karo are two product builders who will be changing the world with products for a long time
Very good practice advice and tips. Thank you