There is a version of AI adoption in hospitality that hotel leaders should resist: the version sold as a shortcut for fewer staff, fewer touches and fewer moments of genuine human connection. It optimizes for the spreadsheet and impoverishes the stay. Guests, more perceptive than we sometimes credit them, notice.
However, when AI is entirely invisible to the guest and deployed thoughtfully, it doesn't have to diminish hospitality. It can help protect it.
The Wrong Question
I co-founded an AI hospitality platform, and when I speak with general managers about AI, the opening question is almost always: "How much can we automate?" I understand the instinct. Labor is the industry's highest controllable cost. Margins are under pressure. The case for efficiency is real.
But "how much can we automate?" is the wrong frame. It treats AI as a replacement technology, something deployed in place of people. For general managers to build genuinely differentiated properties, they must ask a different question: "How much more can our team deliver with AI behind them?" That distinction determines the kind of hotel you build, the culture you sustain and the guests you attract and retain.
The Personalization Gap
Hospitality has always been a knowledge business. The great hoteliers knew their guests: their preferred pillow, the room they liked that's away from the elevator, the anniversary that made a bottle of champagne feel like a memory rather than a gesture.
But scaling that knowledge has always been a challenge. You can train culture; you cannot always train recall across 300 rooms, 12 booking channels and a team rotating across three shifts. AI may be able to help close that gap, not by replacing the instinct of a skilled hospitality professional, but by ensuring that instinct has what it needs to act. For example, AI can be used to surface a guest's profile, booking history and preferences.
Hotels that learn to use AI to genuinely elevate service can build something hard to replicate: a data flywheel. Every interaction can enrich the guest profile. Every enriched profile can enable better service. Better service can drive loyalty and direct bookings.
Hotels that use AI purely to cut costs risk optimizing one variable while degrading others.
What Hospitality Leaders Can Do
For hotel leaders thinking about AI investment, three principles stand out from what we see working across our customer base.
1. Establish the layer that unifies your guest data.
Fragmented systems produce fragmented experiences. The value of AI is directly proportional to the quality of the data it can access. Connecting pre-arrival, in-stay and post-stay information is the foundation everything else is built on.
2. Measure AI by what it enables your team to do, not only by what it eliminates.
If the goal of your implementation is freeing staff to spend more time on guest engagement, and you are tracking that shift, you can build something durable. If you are only measuring cost reduction, you are flying partially blind.
I recommend tracking three things: inquiry resolution rate without staff intervention; response time before and after deployment; and whether staff is spending more time in genuine guest conversations per shift. That last metric is the most revealing.
Keep in mind that meaningful signals typically emerge around 60 to 90 days in. Early performance rarely reflects mature performance.
3. Treat personalization as a brand commitment, not a feature.
I believe the hotels leading this decade will be those that can credibly promise guests: We know you, and we will show it every time you stay. That promise needs to be prioritized when you're adopting AI.
In practice, this can mean connecting your property management system data to your communication layer so every guest interaction draws from the same profile. This also means creating content that's specific enough to actually be sent at the right moment of the guest experience—not just "we offer early check-in" but a workflow that surfaces availability to the right guest at the right time. The hotels doing this well treat guest data as a service asset.
Challenges To Prepare For
The most common failure is not technological. It is incomplete content. AI can only answer questions it has been given material to answer. Deploy without a rich knowledge base, and most queries will escalate back to staff, thus defeating the purpose. The fix is to treat the knowledge base as a living document, review what the AI is being asked, fill the gaps and revisit quarterly.
The second failure is cultural. If staff view AI as a threat rather than support, adoption will be passive. Seeing strong outcomes requires involving your team early; showing them what the AI handles and what it escalates; and framing the shift as giving them back time for the work that actually requires a person.
And one more: AI is not a substitute for staffing judgment. Deploying it to compensate for understaffing rather than augment a capable team produces an experience that feels hollow. Guests notice. That is the version I warned against at the start.
Most hotels manage guest relationships across a patchwork of disconnected strategies. The guest data exists—it is just not working. For leaders willing to think about AI as service infrastructure rather than a cost-cutting exercise, the window to build a meaningful advantage is open. The most powerful thing AI can do for hospitality is make your team better at being human. That is the version worth building.
David Mezuman is the CEO and Co-Founder of Duve.
