Product Judgment: A Conversation with Natalia Klementeva
As AI accelerates how products are built, competitive advantage is shifting form execution to judgment. The real challenge is no longer how to design, but what deserves to be designed.
Natalia Klementeva is a Product Designer with experience building AI-powered products in fast-growing startups. In this interview, she shares why great product design is no longer about interfaces alone, but about shaping the strategic decisions that define what gets built and what doesn’t.
Beyond Product Design
Hello Natalia, thank you for joining us in this new edition of Talent-R Tech Talks, you’ve spent years working in product across AI and e-commerce environments. At what point did you realize that designing a great experience and business strategy could no longer go down separate paths?
“For me they were never separate. A great experience that doesn’t make money is a portfolio piece. A business that ignores experience burns users faster than it acquires them.
The real shift happened when I stopped measuring my work in screens and started measuring it in behavior. Does the user come back? Do they pay? Do they bring someone else? Behind every one of those metrics is a person deciding whether your product respects their time.
Once those become your design metrics, strategy stops being someone else’s job. You can’t answer those questions with UI alone. You need to understand the market, pricing, and the funnel. So I learned them.”
“My failure wasn’t in any single screen. It was that for months I answered “How should this work?” without asking “Should this exist”?”
In many companies, design enters the conversation once the key decisions have already been made. How have you managed to give design a voice in strategy from the very beginning and not just in execution?
“Honestly, I just stopped bringing opinions and started bringing proof.
Every time the team wanted to build something big on an assumption, I’d ask for one or two weeks to test it with real users first. Then I’d come back with recordings and numbers. You can argue with a designer’s opinion.
You can’t really argue with a video of a real person getting lost in your product. If there’s no time to test, I’ll trust my intuition and make the call, but I’ll set up one simple metric to check later if I was right. That’s the whole trick. After you save the team from one expensive mistake this way, you don’t have to fight for a seat at the strategy table. They start inviting you.”
Building What Matters
You work with research and user data as the foundation of your decisions. When user insights point in one direction and business priorities point in another, how do you navigate that tension to reach a decision that works for both?
“First of all, it’s not a fight. Users are usually right about the problem. The business is usually right about what it needs to survive. Most of the time the real question isn’t “users or revenue,” it’s “revenue now or revenue later.”
So instead of arguing, I just show. I test both directions with real users and bring everyone the same evidence. It works almost every time: founders change their minds very fast when they watch a real person struggle with the thing they insisted on. And sometimes the business still picks its priority over the insight. That’s okay. My job isn’t to win every argument. It’s to make sure we decide with open eyes and agree upfront how we’ll know if we were wrong.“
In AI startups like Foxtery, product and technology evolve at an incredible pace. How do you decide what to design first when almost everything feels urgent and resources are limited?
“I’ll tell you how it actually happened at Foxtery. We were starting from zero, and our founder had about twenty ideas, all of them urgent. Classic startup.
So before designing anything, I took two weeks and studied the competitors. What does every product in this market already have, and where do they all fail their users? That map basically made the decision for us. We started with the course builder, the most boring feature on the list. Not because it was exciting, but because without it nothing else in the product made sense. Only after real instructors were creating courses did we start adding our unique features, one at a time, testing each one.
My rule since then is simple: when everything feels urgent, ask what has to work before anything else can. Build that first. Every feature you add before the core is proven is just a bet that the core works. And at a startup, you can’t afford many of those bets.“
Judgment Over Execution
Do you remember a design project that was well validated, well executed, but once it reached users, it didn’t have the expected impact? What changed in the way you work after that experience?
“Yes, and it’s not a comfortable story. We grew our AI product to more than 10,000 monthly active users, with 8% converting to paid. By the numbers, it worked. And the business wanted to build on that success, so new feature requests kept coming. I designed them all properly: research, clean flows, edge cases covered. Each feature, taken alone, was good work.
But the expected impact never came. New features barely moved the metrics that mattered. And when I ran testing sessions to understand why, I saw it: the product had become heavy. Users couldn’t find the core function anymore, because my well-designed features were standing in the way. Nothing I made was bad. All of it together was. My failure wasn’t in any single screen. It was that for months I answered “how should this work?” without asking “should this exist?” I executed well when I should have pushed back early.
What changed: now, before designing anything new, I ask what we’ll remove or measure if it doesn’t perform. And I always test the whole product, not just the new piece. That habit is what eventually gave me the data to walk into the founder’s office and argue for cutting the scope, and he agreed. So the failure taught me the exact skill I’m now known for.”
With AI redefining what can be built, how do you think the role of the Product Designer will evolve over the coming years, which skills will become more relevant, and which ones will be transformed?
“A few months ago I made a bet on Design Engineering: learn to build, close the gap with engineers. Today I design, build and ship real products in Claude Code myself.
But the way it played out surprised me. I expected code skills to become the differentiator. Instead, AI is automating execution much faster than judgment. Anyone can generate screens and even whole products now. What nobody can generate is knowing which problem deserves to be solved and why a user drops at step three. Companies no longer need someone who arranges screens. They need someone who can spot friction, explain trade-offs, connect a UX decision to a retention curve, and sometimes act as a business partner to the founder.
So the skills at the top of my list are critical thinking, knowing the business, and the speed of rebuilding your own skillset without panic. I’ve already had to reinvent mine, and I assume I’ll do it again. The fundamentals stay the same: user experience, research, business metrics, data. And one prediction: the next wave of strong founders will come from design.”
The next generation of successful products won’t be defined by how fast they’re built, but by the quality of the decisions behind them. In a world where anyone can build, knowing what truly deserves to be built becomes the ultimate competitive advantage.
