Why AI Is Putting Hotel Property Technology Architecture Under Pressure

For most of the last decade, the hotel property technology playbook was based on the idea of additions. A new booking engine here, a guest messaging tool there, an upsell platform, a chatbot. Each one sat on top of a property management system that in many cases predated the smartphone. The stack grew taller. The foundation never changed.

That approach is reaching its limit, and AI is the reason it is finally visible.

You cannot personalize with data a model cannot read.

The appetite for AI is not in question. Research from early 2026 found that 82% of properties planned to increase their AI use over the following year. The harder question is what that AI will actually run on. A model asked to anticipate a guest’s needs can only work with the records it can reach, and in a fragmented stack those records are scattered across systems that were never designed to share.

Doug Lange, VP of IT Strategy at Choice Hotels, framed the choice cleanly. He said the brands that treat AI as core infrastructure rather than an add-on will be the ones best positioned to compete, and that Choice is deliberately embedding AI across the enterprise instead of layering point solutions on top of systems that are already disjointed. That last phrase is the whole argument. You cannot layer intelligence on top of disorder and expect intelligence back.

The industry is debating how far to go.

Tan Bee Leng, Chief Commercial Officer at The Ascott Limited, described the two paths honestly. Some players are rebuilding from scratch, others are layering on top of what they have. Ascott’s own approach is to make its systems operate as one unit, so that AI initiatives sit on a foundation that is already clean and consistent. The reduction is not about cutting systems for its own sake. It is about making the ones that remain agree with each other.

Both leaders are describing the same conclusion from different angles. The foundation comes first. The intelligence comes second. Reverse the order and you may just get expensive disappointment.

What a rebuild actually buys you.

The scale of what this involves is easy to underestimate. When one hotel chain completed its move to the cloud, it retired more than 3,700 servers and over 300 applications to get to a single, cloud-first foundation. That is a decision to stop maintaining the past so the future has somewhere to run.

What that foundation buys is the ability to see one version of the guest across every system that touches them, which is the precondition for anything AI is supposed to deliver. Clean, connected data is not the boring part of the AI story. It is the part that determines whether the rest of the story happens at all.

This is where I would push the conversation one step further than it usually goes.

Most of the rebuild discussion focuses on internal systems, the PMS and the CRM and the reservation platform talking to one another inside the property. That matters. But a large share of the guest records are created outside those walls, at the point of distribution, before the guest ever arrives. The channels a property sells through, the content and rates that flow out to them, the booking that flows back. If that layer is fragmented, the internal rebuild inherits messy inputs no matter how modern it is.

A cloud-first foundation and clean distribution are the same project viewed from two ends. One without the other leaves a gap exactly where the guest relationship begins.

This is the layer DerbySoft has spent its business building. We connect properties to hundreds of demand channels and keep content, rates, availability and payment data moving accurately between systems that were built independently. 

Our Property Connector work makes sure detailed, structured property and room information travels intact to every channel a guest might shop, and our AI tools for extracting, reviewing and completing that content exist because at the scale of thousands of properties, accuracy cannot be a manual task. 

That is the input layer a cloud rebuild depends on. 

When companies talk about why they made the move, better connectivity usually comes up pretty quickly. If you want AI working across discovery, booking, revenue management, distribution and pricing, all of those areas need to be working from the same information instead of pulling from different systems and different versions of the truth.

About the Author

Duane Overgaard is the Divisional CEO, Hospitality, of DerbySoft. With over 30 years of experience in the hospitality industry, he has a diverse skill set that includes account management, business development, and contract negotiation.

Learn More About DerbySoft