Home Blogs Sharing judgment, not hoarding it: How one product leader is creating an AI operating system his whole team builds on
  • AI Innovators
  • People of OLX

5 min. read

AI Innovators at OLX

Sharing judgment, not hoarding it: How one product leader is creating an AI operating system his whole team builds on

An interview with Mohit about building the Motors operating system, bringing AI into how the whole org works, sharing how he thinks, and why PMs are becoming builders.

Mohit Kumar
Director of Product, Motors

Mohit, you built something called the “Motors operating system.” What problem were you trying to solve?

We were already running a tightly coordinated multi-functional org — product, design, engineering, analytics, research, sales, marketing, gtm, legal, across Portugal, Poland, Romania. Best-in-class teams, structured ways of working.

But that multi-layer complexity means data and information sits everywhere. Google Drive, Confluence, Jira, Slack threads, handwritten notes. All fragmented.

That fragmentation creates a real cost. Every status report required manual aggregation. Getting a simple answer meant a 1:1 sync or a Slack back-and-forth. People were spending meaningful time on coordination and information gathering instead of building.

And then AI arrived — and everyone started experimenting. Endless individual agents, no visibility into what existed, so each person started from scratch.

That’s when I started thinking: what if there was a shared system that did all of this work, so people could focus on the meaningful stuff. 

Mohit AI Jeremy (2)

You described yourself as a “calibrator” — someone who uses AI not to do more work themselves, but to raise the whole team’s capability. What does that mean?

Being a leader means making sure your team can operate at the highest level — and that includes taking full advantage of the latest technologies. It also means leading by example.

The more I could encode — my thinking on strategy, right outcomes to prioritize, what good OKRs look like, how to write a status report, what a good competitive brief looks like — the more the team could run at that level without needing me in the room.

AI made that scalable. You’re not replacing judgment, you’re distributing it.

Walk us through what the operating system actually does.

The operating system is a shared Claude project that the entire Motors Professionals team works with. It has a shared context layer — who the customer is, who we are, what we’re building, our squads, our markets, our OKRs — and on top of that sits a library of skills built by and for different functions, connected to all the existing tools that we use. 

PMs have skills for status reports, OKR reviews, competitive benchmarking. Design has skills. Analytics has skills. Marketing has skills. Anyone in Motors can use any of them.

If a new PM joined tomorrow, the system could build an onboarding plan specific to that person. They can ask follow-up questions immediately, without waiting for a 1:1. The onboarding becomes personal.

And when they do sit down with existing people, the conversation is about judgment and direction — not “how does this work” or “where do I find this.”

The context that used to take weeks to absorb is just there. The system lets them spend their first week thinking, not onboarding.

What completely failed?

A few things.

Token costs. Every question pulls in full context. And because everyone is still learning to prompt well, it often takes a few tries to get the right output. I now check token usage every month and adjust the setup to cut cost without hurting quality.

Keeping the root context up-to-date was another gap. You’re relying on each function lead to manually update their area. That doesn’t scale. We’ve since partially automated that update — but it took a while to get there.

And in the beginning I was trying to build too many skills myself. I became the bottleneck. What works now is PMs building skills that other PMs can reuse. Design and analytics teams building skills anyone in Motors can access. And same goes for Product marketing and GTM teams. 

The main thing I learned is that a shared system only works when the team owns it, not one person.

What was the reaction when people outside your immediate team started using it?

There wasn’t one big moment, and I think that says more than a single moment would.

Adoption was organic. We told everyone in Motors it existed, and people started trying it at their own pace. Over time I noticed fewer of a certain kind of Slack message — the “where do I find this?” or “can you quickly explain how this works?” ones.

People were asking the system first.

From there each function took it in its own direction. PMs use it as a thought partner and to draft strategy docs. Marketing and GTM builds presentations from context that already lives there. Some people have automated their weekly reporting.

So the moment is small, and it happens every day. Someone opens it, expects a useful answer, and gets one. When that happens across a whole team, it changes how people work.

IMG_0759

What did OLX do that made building this possible?

A lot, and more than people outside might expect.

OLX gives people a real choice of AI tools: Claude, Gemini, OpenAI, automation platforms. It backs that up with training, time to experiment, and a simple process to ask for more capacity when you need it. Not many companies this size do that.

That’s why things like this operating system exist, and why similar ones are showing up across OLX. We could try things, keep what worked, drop what didn’t, without a long approval process.

The executive team has been a strong part of this too. They use AI tools themselves, their direct teams use it, and that signal travels down. When leadership visibly adopts something, it stops being an experiment and starts becoming a culture.

That’s what’s happening at OLX right now.

You mentioned you’re on paternity leave. How does that prove the system works?

I’m currently on paternity leave, and the system is still running.

My team can get their OKRs reviewed using a skill I built — same standards, same feedback quality — without needing me in the room or on a call. And this is just another example of what the operating system can do. 

That’s what it means to distribute your judgment rather than hoard it.

How has your role as a PM changed — and what does it mean that PMs can now build things that used to require engineering teams?

The lines between roles are blurring, and fast.

A year ago, if a PM wanted a dashboard, they filed a ticket and waited. Now a PM with the right data access can query our data warehouse, wrap it in a skill, and share it with the team the same day.

With the right tools, a PM can build a prototype, test it with customers, and then bring in a designer to take it further. We’re even helping PMs build things that actually ship.

That changes what a PM is. You’re not only identifying the right outcomes and writing specs anymore. You’re building.

It works the other way too: a designer or engineering manager with strong product sense can now do a lot of the PM job well.

I don’t think we’ve worked out fully yet what this means for how we hire, how we evaluate people, or how we set up teams. But we are having those conversations and figuring it out together.

What’s your biggest learning from building this?

That transformation isn’t simply about giving everyone the same tool. It’s about building systems that let people work at a higher level.

The operating system isn’t AI doing our jobs. It’s AI removing the friction so we can focus on the meaningful work — strategy, judgment, building.

And that the best test of any system is whether it works when you’re not there.

Right now, I’m not there. And it’s working.

Mohit’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.

At OLX, you don’t just work inside what the company already is. You get to help build what it becomes.

Explore open roles at OLX

Related articles