AI didn’t kill design.
It did something weirder: it made “fine” the default.
Lately I’ve been noticing a side effect of AI and vibecoding tools:
We’re shipping more products that nobody actively **hates**… but nobody really *loves* either.
That’s what this video is about — not AI vs designers, but this uncomfortable question:
If everything is starting to look the same, where does *your* product’s personality actually live?
I’ve been playing around a lot with AI tools lately, and there’s this one setting I kept ignoring because it sounded… technical.
Temperature.
But once I actually understood it, so many weird AI moments suddenly made sense.
But if you nudge it toward 1, it starts taking risks - more creative wording, unexpected ideas, sometimes a little chaos.
And as a PM, that made me rethink things.
We’re not just choosing “accuracy” or “creativity” - we’re choosing the personality of what we build.
I’m starting a new personal challenge: #learninpublic
Instead of waiting until I’ve “figured it all out,” I’ll share what I’m learning while I’m still learning it.
My first video dives into something we’ve all seen (and maybe cursed at): AI hallucinations.
Here’s the line that sparked it:
“ChatGPT doesn’t try to be right. It tries to sound right.”
Once you realize that, it changes how you think about building AI products.
As PMs, the key question isn’t “can we make it accurate?” - it’s “what’s the cost if it’s wrong?”
My startup died.
𝘏𝘢𝘳𝘥 𝘵𝘰 𝘴𝘢𝘺 𝘪𝘵 𝘰𝘶𝘵 𝘭𝘰𝘶𝘥, 𝘦𝘷𝘦𝘯 𝘢 𝘺𝘦𝘢𝘳 (𝘢𝘯𝘥 𝘢 𝘩𝘢𝘭𝘧) 𝘭𝘢𝘵𝘦𝘳.
After years of building my own thing - pitching, hustling, dreaming - I suddenly found myself starting over… this time as a Product Manager at Pipedrive.
It taught me that ownership isn’t about titles - it’s about how much of your heart you put into what you build.
So yes, my startup didn’t make it. But the lessons didn’t die with it.
Merve Cankiz Coruh retweeted
You signed up for a SaaS product.
You couldn't figure out how to do something.
Checked the documentation.
Searched YouTube.
Or took the boring route... contacted support...
These moments are critical for users.
If they solve their problems as expected, no problem.
But if they DON'T...
Product adoption crashes.
Satisfaction decreases.
Lifetime value is at risk.
And churn signals start appearing in your analytics within days.
The question:
How can we support these users during these moments of frustration, instantly?
We built the first "Agentic AI Cursor" that acts on behalf of users. Instantly and 24/7.
Users write their prompt to the AI Cursor, auto-browsing mode starts, and the AI Cursor does everything for users—clicking, typing, moving, searching, and more...
If you want to solve these kinds of moments for your users, contact me via DM.
Merve Cankiz Coruh retweeted
AI dünyasında en son çıkan yenilikleri takip etme konusunda öncü olan arkadaşlarınızı bu postun altına etiketleyebilir misiniz?
Bu arkadaşları @mues_ai ‘ı ilk deneyimlenenlerden olmaları için birebir görüşmelere davet ediyor olacağım.
Ayrıca kendinizi mentionlamak serbest 🪽