The PMS is dead. Long live the PMS.

Welcome to The Night Porter. We start where every good argument does: by telling the smart person they used the wrong word.

A few weeks ago Adam Harris, the CEO of Cloudbeds, published a piece that did very well, declaring, from the conference in Boston, and then to a few hundred thousand poeple on LinkedIn tha “ PMS is dead. Nobody on any conference floor has had the stomach to call it “

It’s a good piece of writing and I want to agree with the true parts before I take the wrong part apart, because the disagreement is only interesting once agreement is on the table.

He is on point about legacy property management systems, “the filing cabinet… built 25 years ago” is finished. He is right that hospitality data is scattered across systems that cannot see each other, let alone communicate with each other. He’s, right, most of all, abou that borrowed line of hotelier Wagner Quintanilha: hospitality is “ someone that remembers “.

When s king dies, the announcement isn’t “the throne is empty”, but “The King is dead! Long live the King!” This is said in one breath because the heir is already sovereign. Correct? Correct. The institution is continuous, only the occupant changes.

That’s the PMS. The old one is dying. The thing replacing it, one system, one guest record, one model is the next PMS? Adam isn’t ending a category, he is crowning its successor and calling it a funeral, because funeral is better keynote. Because you, reader, don’t have to mistake a coronation for an autopsy. And the moment you call it a succession, a more useful story conjures.

I’ve read Adam’s words carefully and…

Take the hospitality nouns out of the article and you’re left with he oldest complaint in enterprise software:

our customer data is trapped in disconnected systems, so we can’t act as one business toward one customer

I’d argue silos are the single most expensive unsolved-then-solved problem in the history of business software and the industry that’s about to discover them has a quarter-century of other people’s homework to copy (and hopefully improve upon).

Here we go:

1. Single Source of Truth — the data-warehouse promise of the 1990s.
2. Master Data Management — he “golden record”, the “Customer 360”, the 2000s’ entire answer to “which system is right about this customer?”
3. Customer Data Platform — a category named in 2013 to do precisely what Adam describes: unify customer data into one model that downstream tools (today, AI) can act on.
4. System of Record → System of Engagement — Geoffrey Moore’s 2011 framework (the “system of intelligence” layer got bolted on a few years later), which is, almost to the word, the layering the smartest commenters on Adam’s aricle reach for without the vocabulary.

And here is the tell that should make any veteran smile. Every time IT runs this play, someone declared the incumbent dead.

The mainframe is dead. The relational database is dead. The website is dead. Email is dead. — said everyone for the past quarter century.

The truest sentence in the whole hotel-tech conversation said exactly this, and it came from a commenter, not the keynote. Michael Levinson, who built one of the first online hotel reservation platforms in 1996: “ The tech caught up 30 years ago. The industry hasn’t”.

If the PMS were really dead, the most advanced operators would have walked away from it. They’ve done the opposite.

So why fight over a verb? Because the verb hides the real questions, and the real questions is about the power The pitch is freedom: own your guest data and

“the chain doesn’t have a monopoly… You do.”

(What and odd way to phrase this! I wonder why. I am absolutely sure Adam knows what the monopoly is. And I don’t believe for a second he used monopoly as a mistake or a rhetorical device)

But trace the mechanism. You “own the guest” by pouring every guest’s data into one unified model, and whoever owns that model is sitting in the intermediary’s chair. Booking.com disintermediated hotels by owning the demand. The owner of the unified guest record can re-intermediate them by owning the memory. Gur B. said it best in the comments: “the question isn’t what it remembers. It’s who the memory serves”. Kirill Lebedev named the endgame, that today’s “hospitality OS” can become “just another middleware tier waiting for its own funeral on stage” the moment agents can read any system on their own. The logic that kills the old king eventually knocks on the new king’s door. That’s the pattern. It never stops at one coronation.

Often the most useful thing I can tell this industry is that it isn’t as alone, or as new, as the keynote makes it feel. Other people have stood exactly here before, in other lobbies, under other acronyms. The PMS isn’t dead; it’s being crowned again. The only question that matters is who it serves once the crown is on.

While the conference stage argues about whether the PMS is dead, the people quietly winning the actual war aren’t selling slogans, they’re selling integration. Look at Wonderful, an enterprise-AI agent platform founded only in 2025 by Bar Winkler and Roey Lalazar, already deploying production-grade agents across more than thirty countries in telecom, financial services, manufacturing and healthcare. Its entire thesis is a rebuke to the “costs pennies, one superhuman employee by Friday” fantasy:

“Enterprise AI will not scale through technology alone. It requires a state-of-the-art agentic platform paired with locally embedded teams that can deploy agents inside complex organizations”.

The model, in other words, was never the hard part. The hard part is the two failures every IT veteran already knows by heart:

1. the demo-to-production gap, the chasm where flashy pilots quietly die of hallucinations, broken legacy systems and
2. local compliance quirks, and what Wonderful calls the “AI deserts”, the way US-centric models stumble the instant they hit another language, another culture, another regulator.

And the mechanism is the part that should interest a hotelier, because it maps precisely onto what an operator should want. Wonderful runs a model-agnostic, horizontal architecture, one underlying foundation across an enterprise’s core systems, continuously benchmarking and swapping in whichever model performs best. They they refuse to rebuild the silo.

Then they do the thing self-serve software won’t: they send full-stack engineering teams into the client’s environment to own the rollout, the integration and the post-launch tuning. Because every deployment sits on that one shared foundation, each new one is faster and cheaper than the last: more than 70% of their enterprise clients, by the company’s own account, expand into adjacent workflows within three months of going live. That is saying alot about eliminating silos Adam has spoken about. Connecting the dots is not that difficult.

Read that number as the market pricing the real lesson of every silo war that came before: the winner is never whoever announces the incumbent’s death loudest, but whoever does the unglamorous integration nobody claps for. For hospitality the translation is exact. A hotel is multilingual, multi-jurisdiction, GDPR-bound and welded to legacy systems, an “AI desert” by Wonderful’s own definition. Whoever unifies that guest record and makes it actually work at the front desk at 3am is who wins. You can try to become an architecturally critical player and act mala fide to kill the PMS, but PMS won’t die of it’s own volition. Why would it? Why would you even try to do it? To disrupt for the sake of disruption? Nah. Which lands us, one last time, on the only question that ever mattered: once the integration is done and the memory finally works, who does it serve?

That’s what you’ll get here: hospitality told straight, from the quiet hours: tech and AI included, none of it taken on faith just because someone said it from a stage. If that’s your kind of thing, subscribe. The next king is always being crowned somewhere down the hall, and somebody’s always calling it a funeral.

The PMS is dead. Long live the PMS.

Intellectual Lineage

Adam Hariss LinkedIn.com Post, Geoffrey Moore’s 2001 Escape Velocity: Free Your Company’s Future from the Pull of the Past, Insight Partners · TechCrunch — $150M Series B at $2B valuation · Calcalist/CTech · Index Ventures · EU-Startups (Amsterdam HQ, €1.7B)

Originally published at https://darioalfirevic.substack.com.