Klook launches Kai, an AI agent that runs merchant back-office tasks

Klook has launched Kai, an AI agent for the merchants selling experiences on its platform, framing it as an operator that acts rather than an assistant that only answers.

According to Klook, Kai runs inside the merchant portal and can carry out inventory management, pricing adjustments and booking updates. Those are exactly the repetitive chores that dominate a small operator's day: inventory has to move with seasonality and weather, prices have to follow demand, and booking changes have to be written back into the system quickly.

Klook said Kai will roll out globally in phases ahead of the year-end travel season. The first target is small and medium-sized businesses, which Klook says make up roughly 95% of the experiences sector and often lack the digital infrastructure to sell beyond offline channels.

Bernie Xiong, Klook's co-founder and chief technology officer, said the company wants to bring its most practical agentic capabilities into merchants' day-to-day operations and remove much of the manual work involved in running their businesses.

Klook also published early-adopter figures: Kai completes some tasks up to 97% faster than traditional processes, and merchants using the agent recorded 34 times more monthly bookings than those who did not. The numbers come from the platform's own sample with no published methodology, so they should be read as vendor-reported rather than independently verified.

Beyond routine operations, Kai simplifies product listings and analyses customer reviews and performance data to surface suggestions merchants can act on.

The larger signal is that travel platforms are competing on back-office supply as much as on front-end demand. As traffic growth matures, automating the operating layer for small merchants becomes a way to lock in both efficiency and loyalty.

What to watch next: how far the phased rollout actually reaches, whether usage holds through peak season, and whether Klook publishes audited metrics rather than launch-week comparisons.