Automating 5,000+ Monthly Guest Requests for a Serviced Apartment Operator

Every booking at HavenStay carried a hidden cost: the flood of routine guest requests that followed it. Parking questions, early check-ins, invoice copies, booking extensions — the same handful of queries arrived thousands of times a month across 90+ cities.

The support team was answering identical questions hundreds of times a day, while complex guest situations that needed real human judgement sat in the queue. The finance team alone was fielding over 1,000 invoice requests per month by hand.

The conventional fix — hire more agents and scale support linearly with bookings — would have eroded margins with every new city launched.

Instead of a risky big-bang rollout, MintoraX proposed a focused pilot: automate guest invoice requests first, prove the accuracy, then expand.

The pilot agent hit 94% accuracy within weeks. From there we designed and shipped a full guest-operations suite of six specialised AI agents:

• Check-in Agent — online check-in and ID queries
• Check-out & Extension Agent — stay extensions and late departures
• Parking & Luggage Agent — the single biggest volume driver
• Early Arrival Agent — arrival-time coordination
• Group Booking Agent — multi-unit enquiries
• Invoice Agent — automated invoice generation and delivery

Each agent plugs into HavenStay’s booking platform and messaging channels, with human handover for anything ambiguous. We also trained HavenStay’s own teams to build and iterate on agents themselves — no ongoing engineering dependency.

Six agents now handle over 5,000 guest interactions monthly. A third of all service requests — parking, early arrival, extras — are resolved end-to-end with zero human involvement, and routine response times dropped from minutes to seconds.

The support team now spends its time on the situations where a human actually matters, and two additional internal teams are onboarding onto the platform to automate their own workflows.