Home Blogs “Put the machine to do the repetitive work you don’t like. Then use it to do the things that generate impact and you love even better.”
  • AI Innovators
  • People of OLX

5 min. read

AI Innovators at OLX

“Put the machine to do the repetitive work you don’t like. Then use it to do the things that generate impact and you love even better.”

An interview with Leonardo Eloi, Senior Product Manager at OLX Motors — on living on a boat, nearly giving up on AI, and why he built an agent to sell his own stuff on OLX.

Leonardo Eloi
Senior Product Manager at OLX Motors

Leonardo, let’s start with the basics — who are you and what do you do at OLX?

I’m a Product Manager in the Motors tribe, working with professional dealerships. My focus is on creating a seamless experience for large, complex accounts — think a dealer group like Plichta in Poland, with over 70 dealerships, multiple brands, and a fragmented set of separate accounts. My job is to connect all of that. To make something that feels messy and manual feel unified and simple. Technically it’s very complex, product-wise as well — but I love that. When you solve one problem, there is always a risk you create another. And most of the time it is ok as long your goal is always to have a net positive impact.

LinkedIn AI Leonardo (2)

Before OLX, you were at Volkswagen. What brought you here?

I was working on a project connected to onboarding new buyers of the electric ID car family when the pandemic hit and the supply chain went into shock. No cars to make. I started thinking about used cars — helping dealerships sell smarter, helping buyers make better decisions. I’d had a bad experience myself: I bought a convertible in Portugal that leaked every time it rained. I tried to pitch a used car initiative internally at Volkswagen, but it was taking forever to move. Then OLX reached out and said: we’re building exactly that. I thought — oh. This is interesting. And it was the right fit at exactly the right moment.

There’s also a less conventional part of your story — you once lived on a boat?

Yes! Early in my career in Brazil, I had a long commute to an office that happened to be right next to a marina. I thought: why not? I found a cheap old boat, fixed it up, and lived on it for almost two years.

It was minimalist by necessity — tiny space, limited 3G internet, just me and my computer. But it taught me a lot. How to build things that are simple and functional but also genuinely enjoyable. How to optimise when resources are limited. How to reflect, because I had a lot of time alone with my thoughts. I think a lot of my product instincts were formed on that boat.

(And that’s why I only did it for two years, not a lifetime.)

Let’s talk AI. You committed to learning it as part of your development plan back in 2023. But it didn’t go smoothly at first?

No, not at all. The tools were genuinely not there yet. I remember trying to use AI for competitive research — spending more time figuring out how to prompt it than if I’d just done the research myself. The outputs weren’t great. I knew enough from experience to see when it was wrong. And I thought: I’ll just do it myself.

But I put it in my calendar. I kept coming back. I studied prompt engineering. I watched the models improve. And slowly, I started to see: okay, this is getting better. But I still had moments of real frustration — even recently. The difference now is that when something doesn’t work, I ask myself: what did I do wrong? Did I give it too little or too much or wrong context? Did I provide the wrong file? Not enough detail? I’m still learning. Everyone is.

So what changed? What made it click?

The agentic shift. Around 2025, it stopped being about talking to AI and started being about AI that understands your context, executes on your behalf, and learns how you work. That’s when everything changed for me. Not just faster — fundamentally different.

Leo pic1

What does your working day actually look like now?

Every morning, my meetings from the previous day are already synthesised — action items, priorities, decisions made, what needs to happen next. That used to take me 30 to 45 minutes at the end of every day. Now it’s just there.

My KPI updates — pulling data from Tableau across multiple countries, calculating deltas, formatting the message, posting to Slack — used to take 20 to 30 minutes every few days. Now it’s one command and seconds.

User stories that used to mean cross-referencing Figma files, PRDs, and meeting notes for hours? I feed in the context and they come back structured and ready for Jira. I review the 5% that needs a human eye. That’s my job now.

And when I need content across markets — Polish, Portuguese, Romanian — simultaneously, in minutes. Not hours.

You mentioned something outside of work too — an agent you built for yourself?

Yes! I was doing a spring clean and wanted to sell some things on OLX. So I built a personal agent — running directly on the platform — to handle it for me. I gave it the rules: what to accept, what to decline, how to describe the items, what tone to use. And it goes and answers buyers on my behalf.

What surprised me was how non-deterministic it is. I just gave it context and rules, and it figured out the rest. At one point, a buyer asked about the storage size of a Meta Quest I was selling. I hadn’t entered that detail. The agent answered 128GB — and it was right. It still surprises me. Because I build the product professionally — and I’m using it personally, in exactly the way it’s meant to be used.

What would you say to someone who’s curious about AI but doesn’t know where to start?

Be curious. And be disciplined — because you’re not going to have your aha moment on day zero. It takes time. You’ll get frustrated. That’s part of it.

Start with something that genuinely annoys you. Something repetitive that you don’t enjoy that much, that takes time, and that doesn’t require your best thinking. When you see that friction disappear — your brain clicks. And then you start to understand what’s actually possible.

Because here’s the thing: when we do work we love, we do it better. We excel. So the goal is to put the machine to do the things you don’t like — and use that freed-up time to do more of what you’re actually great at. Then, later, use AI to help you do those things you love even better.

It’s not about replacing what you do. It’s about getting more of the best of it.

Leo’ s 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.

Think you’re ready to build, not just imagine? Join us.

Explore open roles at OLX

Related articles