Carousell CEO on How AI is Raising Trust Across the Marketplace

When you open the Carousell app, the first thing you notice is the flood of listings – everything from vintage sneakers to second‑hand furniture. It’s a bustling digital bazaar, but behind that vibrant surface lies a serious challenge: how do you keep the experience trustworthy when strangers are transacting with each other?

In a recent interview, the Carousell CEO, Quek Siu Rui, walked us through the company’s latest playbook. He said the answer isn’t a single magic button, but a suite of AI tools that work silently in the background, spotting risky behaviour before it harms a user.

“We’ve always believed that trust is the currency of a marketplace,” Quek explained, pausing for a moment as if recalling the early days of the startup. “Back then, we relied on community reporting and manual reviews. That was fine when the platform was small, but today we’re handling millions of transactions daily. Human eyes simply can’t keep up.”

Enter machine learning. Carousell now runs a series of models that evaluate everything from the language in a listing’s description to the pattern of a seller’s activity. One algorithm flags unusually high‑priced items that match known counterfeit descriptions, while another watches for rapid posting bursts that often precede phishing scams.

These systems aren’t perfect – they’re deliberately designed to be a little cautious. “We prefer a false positive over a false negative,” the CEO said with a chuckle. “If a legitimate seller gets temporarily blocked, it’s an inconvenience we can fix quickly. If a scammer slips through, the damage is far bigger.”

But AI isn’t just about catching the bad actors. Carousell has also rolled out a verification badge powered by facial recognition and document analysis. Sellers who opt in can have their identity confirmed, and the badge appears next to their name, giving buyers a visual cue that the account has been vetted.

“People feel more comfortable when they see a ‘verified’ label,” Quek noted, tapping his pen on the table. “It’s a small psychological boost that can be the difference between a sale and a missed opportunity.”

The company’s data team constantly refines the models using feedback loops. When a user reports a listing as suspicious, that information feeds back into the training set, making the AI smarter over time. The CEO emphasized that the technology is always learning, adapting to new fraud tactics the moment they appear.

There’s also a human‑in‑the‑loop component. Whenever an AI flag reaches a certain confidence threshold, a dedicated trust team reviews the case. “It’s a partnership,” Quek said. “AI handles volume; people handle nuance.”

Beyond safety, the AI initiatives are having a surprisingly positive side effect on engagement. Verified sellers tend to see higher conversion rates, and buyers are staying longer on the app because they feel more at ease exploring listings. The CEO pointed to a recent internal study showing a 12% uplift in repeat purchases after the verification feature launched.

Looking ahead, Carousell is experimenting with predictive pricing tools that suggest optimal prices based on market trends, again using AI. While still in beta, the hope is that smarter pricing will reduce the temptation to overprice or undercut, which can sometimes lead to disputes.

In sum, the message was clear: AI isn’t a silver bullet, but it’s becoming the backbone of Carousell’s trust engine. By blending sophisticated algorithms with human judgment, the platform aims to keep the marketplace lively, safe, and, above all, reliable for the next million‑plus users joining every day.