The question isn't whether AI can do something. It's whether you can trust it to do it at your property, with your guests, to your standards. A wrong rate published across channels costs real revenue. A VIP preference ignored costs a relationship. These aren't edge cases – they're Tuesday.
Part 2 of Rogers Leo's series on agentic AI tackles the trust question head on.
How Mews thinks about AI autonomy
Every agent Mews builds starts with human-in-the-loop oversight. Agents propose. Humans approve. This isn't a temporary training wheel – it's a design principle called earned autonomy. Agents demonstrate reliability before their scope of independent action expands. You don't hand decisions over; you let the system show its working first.
Transparency is what makes this work in practice. When a front desk agent can see that a room was suggested because it matched a guest's historical preferences and avoided a housekeeping conflict, they can evaluate and challenge that recommendation with confidence. Staff don't need to understand how the model works. They need to see that the reasoning makes sense in a context they know better than anyone.
The system gets smarter every time you correct it
Every override, approval and correction is fed back in. A property that has been running an agent for six months has a fundamentally better system than one that just switched it on. Institutional knowledge isn't replaced – it's encoded. The front desk manager who knows Mr. Chen always wants the corner room on the sixth floor doesn't get overridden by an algorithm. That preference becomes part of how the system reasons for every future visit.
For premium properties, the bar goes further than functional accuracy. An agent that produces the right outcome in the wrong tone – too casual for a luxury brand, too formal for a lifestyle property – has still failed. Mews evaluation frameworks account for tone, taste and brand consistency, not just whether the answer was correct.
(Part 1: Beyond the Chatbot — The Agentic Future of Hospitality | Mews Community)