5 min. read
An interview with Inês Urbano, Senior Product Designer at OLX — on building a product feature in 7 days, learning to ask bolder questions, and why speed is only half the story.
Inês, set the scene — what was the 7-day project actually about?
Our team — product, design, and engineering — had a challenge: could we build a real, customer-facing product feature in just seven business days using AI agents? The feature itself was a Tinder-style property discovery experience for OLX Real Estate. Users could swipe right on apartments they liked, swipe left on ones they didn’t, and the system would gradually learn their preferences — generating personalised search filters and connecting them seamlessly to the full listing experience.
It sounds ambitious. And it was. But that was the point.
What made it possible was that OLX actually gave us the space to try. FrAIday — our monthly company-wide session for AI experimentation — wasn’t just a demo event. It was a signal. Leadership was saying: we want to see what you build, and we’ll make time for you to build it. A dedicated week, a cross-functional team, the right tools, and a genuine expectation that we’d take risks — that’s an environment, not just a good idea. Without it, this project doesn’t happen.
Has the feature reached real users yet — and what has it taught you so far?
At the time of FrAIday — four hours of live demos, real problem-solving, and a fireside chat with one of Europe’s leading AI thinkers — we hadn’t yet launched. It has since gone into production, so the real customer insights are still arriving. But that’s actually the point.
What the project already demonstrated is how dramatically AI compresses the time between having an idea and learning from the people you’re building for. Before, a project built over a quarter delivered real user insights at the very end — you spent months working on assumptions, and only found out at the last moment whether any of it was right. Now that cycle is completely different. We shipped in seven days. The feedback that used to take three months to arrive will arrive in days.
That changes not just how fast you build — but what you dare to try in the first place.
The questions your team was asking shifted during the week. How did that happen?
That’s one of the most interesting things I took away from this project. And I don’t think it would have happened the same way somewhere else.
At the beginning, our questions reflected the reality we were used to — building and testing ideas was expensive, being wrong had a high cost. So naturally we’d learned to reduce risk before taking action. Our early questions were: “What can we realistically achieve in one week?” “What is essential?” “What can we simplify?”
But as we gained confidence and saw the potential of building with AI, the questions changed. They became bolder: “How can we make this exceptional?” “What would create even more value?” “What if we pushed the limits?”
That shift didn’t happen in spite of our environment — it happened because of it. When you’re working somewhere that genuinely expects you to push, your own expectations of yourself change. The permission was real. That matters more than people realise.
The products we build are ultimately a reflection of the questions we ask — and their quality is often limited by the quality of our thinking. As the cost of implementation decreases, the value of asking great questions only increases.
Did you try something during the 7 days that you would never have proposed in a normal project cycle?
Absolutely. Our original scope was actually much simpler — just the swipe mechanic, letting the system learn user preferences over time. But halfway through the week, we started seeing untapped potential. We asked: “What if, instead of just swiping, users could instantly contact the seller when they found a property they liked? And what if they could save it automatically too?”
The next day, both features were implemented.
That is the complete opposite of how traditional product cycles work. Normally, moving from an idea to production takes months. Once a sprint starts, requirements are often treated as fixed to protect delivery timelines. AI fundamentally changes that dynamic. Instead of separating thinking from building, the two happen together. You can learn, rethink, improve and implement almost in real time.
For the first time, it feels like creativity is no longer constrained by the cost of execution. And I think that’s incredibly exciting.
What was the hardest part? The moment you weren’t sure it was going to work?
Honestly, our biggest concern wasn’t speed. We knew we could move fast and, if needed, reduce scope to meet the deadline. Our real question was different: in the short time we had, were we building something that was actually good?
AI makes it incredibly easy to generate interfaces, code and prototypes. The temptation is to look at the first version, be fascinated by how fast it appeared, and think: “This is good enough.” But that’s often where the real work begins.
We spent a lot of extra hours continuously challenging, tweaking and simplifying our own solution — uncovering its gaps. Because if you build something quickly but nobody understands it or finds it valuable, all you’ve done is create output — not impact.
That’s one of the biggest lessons of the AI era. And it’s something OLX is actively thinking about at a company level — not just how to move fast, but how to build a culture that keeps asking whether what we’re building is genuinely valuable. Speed is easy to celebrate. Quality takes more courage.
Could other teams do this? Is it replicable?
Yes — absolutely. But a few ingredients matter, and I want to be honest: OLX provided them.
A truly cross-functional team. We had product, design and engineering working side by side from day one — that’s how OLX structures this kind of work, and it makes a real difference. Each of us challenged the solution from a different angle, which helped us move quickly without compromising on quality.
The right environment. Seven days fully dedicated to the challenge, access to the right tools, short daily syncs to continuously learn and adapt. That’s not an accident — someone made the decision that this was worth the company’s time and energy. That decision signals something important about what’s valued here.
And a culture where ideas could be challenged without ego. Nobody was defending the first solution. We were all trying to find the best one. I’d like to think that’s who OLX hires for — people who want to find the right answer, not protect their own.
AI gives teams incredible speed. But OLX is trying to build something harder than speed: a culture where that speed gets pointed at the right questions. That’s what I think makes the difference. And it’s what makes me glad to be here while it’s still being built — because the people here aren’t just working inside what OLX already is. We’re helping shape what it becomes.
Inês story is part of our “AI Innovators” series. Across OLX, people are finding new ways to use AI — to create, simplify, learn, and move faster. This series celebrates the colleagues who are sharing their discoveries and inspiring others to experiment along the way.
