“Earn AI builds on top of whatever systems a landlord already has” – Rentify’s Rajneel Kumar on Earn AI

Seven or eight years ago, Rajneel Kumar moved to Dubai, rented an apartment, and hit the same friction that Rentify now exists to remove. He wrote out four post-dated cheques by hand, redid them because his handwriting kept coming out wrong, then hand-delivered them to his landlord, the chairman of a well-known advertising company, in person. “Our first interaction was a transaction,” he says now. “It stayed with me.”

That memory is the plain, human version of a number Rentify now quotes to the press: UAE landlords may be leaving 8 to 14% of their annual rental income unrealised, through pricing gaps, tenant churn and payment leakage. Kumar spent two decades before Rentify building consumer technology at scale, including a role helping grow VOOT into one of India’s largest OTT platforms, and co-founding Coto, an AI-driven platform for women. We talked to him about what Earn AI, Rentify’s landlord-facing AI layer, actually does that Rentify Pay didn’t already do, the real accuracy rate behind its automation, and why he thinks the best way to get landlords to adopt new technology is to avoid asking them to change anything at all.

The Interview

500 million users across gaming and streaming, VOOT, ZEE5. What pulled you into real estate, or proptech? How did that pivot from digital media to building Rentify actually happen?

Rajneel Kumar: I’ve always had a passion for solving consumer problems at scale. The 500 million may have been across different types of businesses, but at the core of it was always a friction the user had that needed solving; that’s really where my passion lies.

I just came from a meeting that ran an hour and forty-five minutes with a large enterprise customer, a head of legal who’d been there 34 years, a leasing manager there for 15. In that room, I turned into a schoolboy who’s eager to learn, trying to understand the actual pain points that exist inside large organisations and what role we can play in solving them. That’s the same passion I brought to the problem I hit personally when I moved to Dubai seven or eight years ago and started renting an apartment.

I came face to face with the friction that exists after you’ve actually liked the apartment. I was renting from the chairman of a well-known advertising company, a very professional, very good landlord. I wrote out four cheques by hand; my handwriting was bad, I had to redo it about six times before it looked right, put them in an envelope, and went to his apartment to hand them over in person. We had to set up time outside both our busy schedules just for that. At every level, it felt unnecessary. I’d rather have gone and had a coffee with him and talked about the city, since I was new there and could have used his advice. Instead, our first real interaction was a transaction. That stayed with me.

Two years back, I met Rashed, who was looking at the exact same problem from the landlord’s side, and he said something needs to be done about this. Whether it was gaming, streaming, health tech, or now renting, the starting point has always been the same: it’s a consumer problem, and my role is to try and solve it at scale.

Rentify’s own numbers suggest landlords are leaving 8 to 14% of their annual rental income unrealised, from pricing gaps, tenant churn and payment leakage. Rentify Pay has already been handling collections for a while. What does Earn AI’s automation actually add on top of that? Where’s the new AI capability versus repackaging infrastructure you’d already built?

Rajneel Kumar: Let me give you an example that’s easier to relate to. If you want to go from here to another city, you get a car, and the car can take you anywhere. But if I told you I wanted you to go specifically to a particular restaurant in a particular neighbourhood, you’d need something else to get you there: a map. The map tells you how long it’ll take, where the petrol stations are along the way.

With Rentify Pay, what we built is a system that lets transactions happen effectively, the car. With Earn AI, we’ve connected the dots between landlord and tenant to make the whole transaction as seamless as possible for both sides, the map.

Go back to my own story: one landlord, one experience, and it was already unnecessary friction. Now imagine that same landlord or property manager has a hundred units instead of one. A hundred cheques. Every payment due date, every renewal, every piece of communication, every request for flexibility. The complexity multiplies to the point where it’s no longer a human problem or even a software problem; it’s an AI problem.

Our job isn’t just to get the transaction done. It’s to make it as delightful and frictionless as possible, avoid the routes where you know there’s going to be an issue, and get people there in great shape. With Earn AI, the goal isn’t only making the transaction or the renewal happen; it’s reducing the cognitive load people carry, remembering to send the renewal agreement, not missing a deadline, offloading all of that to a system that handles it while keeping them in control. So they can focus on what actually matters, like a landlord welcoming a new tenant to Dubai and telling them where to eat, instead of chasing cheques.

Earn AI has features like early-departure prediction and renewal tracking. What’s the model actually trained on, and how accurate has it been in practice? Is there a concrete example where it’s caught something a property manager would have completely missed?

Rajneel Kumar: One of the core things we’ve solved for is an intelligence engine that takes data in whatever form or shape it exists with a landlord, and displays it accurately and actionably inside the system. The model is trained on a mix of simulated data and real data that’s come in, and we’re currently at about 94% accuracy.

The remaining 6% gets flagged, not guessed at. If there’s an “Apex Corporation” and an “Apex Corporation LLC” in the data, it’s probably the same entity, but the system isn’t certain, so it flags it for a human to confirm. A lot of that 6% comes from the fact that there’s no standardised way landlords keep their records; building and unit numbers might be in separate columns, or the same column; tenant names on one sheet and unit numbers on another. The system gets smarter with every case we solve for, and we’re constantly working to improve that number and not let it slip backwards.

One of the clearest examples of revenue leakage we caught was around team transitions. Portfolio managers change, and a lot of institutional knowledge lives only in people’s heads, so it doesn’t transfer cleanly to whoever comes in next. In one real case, there was an upcoming renewal where the previous property manager had verbally committed to renewing at a certain price because of the relationship they had with the tenant. When new management came in, they had no record of that commitment, looked at the number, and said no. The tenant, who’d been there for years, left; understandably, their situation was that the landlord’s management had changed, not them. The unit then sat vacant for two and a half months before it found a new tenant. That entire loss came from a verbal agreement that was never recorded anywhere. A system that captures and preserves that kind of information, even through normal staff transitions, closes exactly that gap.

Rentify sits on both sides of this. Rentify Pay serves tenants, Earn AI serves landlords. How do you handle it when a landlord wants aggressive collection and a tenant wants flexibility? Whose interest wins?

Rajneel Kumar: The simple answer is nobody should win, and everybody should win. If we always make one side win, we haven’t done a good enough job, because a one-sided win is really a defeat for the other person who never agreed to it.

On flexibility specifically, the question from day one was how do we offer it at scale, and at the lowest possible price for the user, because any price increase on a rental hurts; we all feel that as consumers. That’s why Rentify Pay became a fully embedded finance product, with pricing linked directly to the banking system rather than added on top. That lets us offer flexibility at the lowest rate in the market, while the landlord still receives their money upfront.

So picture a tenant who wants flexibility and is willing to pay a little more for it, as long as it’s reasonable, while the landlord still gets what they want: payment upfront, on their terms. That’s both sides winning, and it’s been the objective since we started building the payment infrastructure and banking relationships two years ago. We now work with five or six banking partners actively on the platform, which is what makes it possible to offer that flexibility to landlords on terms tenants are comfortable with.

Real estate isn’t exactly known for moving fast on new technology. What’s been the actual landlord resistance to handing this over to an AI agent, not “landlords are excited about AI,” the real pushback, and how have you addressed it?

Rajneel Kumar: First, there isn’t really a handoff. Earn AI builds on top of whatever systems a landlord already has, whether that’s an ERP or no ERP at all and just spreadsheets and documents stored on someone’s computer. That’s actually where the resistance starts: resistance comes from enforcing change.

This is where my background in consumer behaviour comes in. The best way to change people isn’t to try to change them directly. Think about any couple where one person is trying to change the other; it usually doesn’t work, or it causes conflict, because change isn’t something people want imposed on them. We all want to wake up early and go for a run at six, but the television at two in the morning is more interesting in the moment.

So our belief is that the best form of change is no change. That means doing the hard work of mirroring a landlord’s existing workflow rather than forcing a new one on them, then running on top of their systems to enhance operations instead of replacing them. That’s why adoption has been strong; we’re standardised at one level, but fully customised for how each landlord actually operates, down to the emails they send, how they input data, and how a digital payment notification talks to their internal systems so nobody mistakenly deposits a cheque that’s already been paid digitally. Once you’ve mapped and solved all of that, you can honestly say you’re not making anyone change; you’re enhancing their system with technology that takes work off their plate instead of adding to it.

What’s the one thing Rentify is building or targeting over the next 12 months?

Rajneel Kumar: There are so many features in the pipeline it’s hard to pick one, they’re all things we’re excited about. But if I think about it in terms of the actual problem to solve, tenants want to stay in units and with landlords they like, and landlords want to retain good tenants, because nobody wants to move just for the sake of moving. A landlord’s happiest outcome is a unit that’s never vacant.

What sits between that ideal and reality is payments and payment terms, maintenance, and renewal communication. So looking at the next twelve months, the core product I want to build, and haven’t built yet, is something that understands a tenant’s full profile and what the landlord wants from that unit, whether that’s pricing or retention, and creates common ground for both, so that nobody needs to leave an apartment they love, at a price that works, with maintenance actually handled and nothing left leaking.

I don’t know yet exactly what that feature looks like. But I know the problems are identified, and as we keep shipping pieces that make property managers’ jobs easier and solve each friction point on the tenant side, we’ll arrive at the feature that brings all of it together into that kind of relationship.

Editor’s Take: Neel doesn’t sell Earn AI as an AI story first. His sharpest answer, “the best form of change is no change,” is an adoption and change-management argument, and it’s a more convincing answer to real estate’s slow-moving reputation than ‘landlords love AI’ would have been. His answer on the next 12 months is honestly framed as a problem statement, worth checking back on once there’s an actual feature to point to.