An LLM Answers the Question You Asked. A PHAL CEO Wants Hotels Ready for the One You Didn't.

PHAL CEO Saurabh Prakash draws a sharp line most hospitality AI coverage skips: “An LLM answers the question you asked. AGI will answer the question you did not know to ask.” His argument isn’t that hotels need to deploy AGI today — it’s that the data architecture decisions properties make now determine whether they can use it when it arrives. His vehicle for that is a framework unifying four metrics typically owned by four separate, non-communicating teams: Performance, Profit, Preference, and People. As long as those four dimensions live in siloed systems and separate reporting lines, Prakash argues, no AI system — however capable — can act on them coherently, because it never sees the whole picture any single department does.

That’s a variation on an argument Hospitality Net ran three weeks earlier from a different angle. Pertlink’s Terence Ronson mapped hotel tech as a four-layer stack — system of record (the PMS), an application/semantic layer, the frontier model, and compute — and warned that AI lock-in is migrating one layer up from the old PMS lock-in fight: whoever owns the semantic layer that translates raw hotel data into something an agent can act on effectively owns the guest relationship. Ronson’s prescription was model-agnostic contracts and auditing that semantic layer now. Prakash’s is functionally the same instruction pointed at a different silo problem — unify Performance, Profit, Preference, and People before you need a system smart enough to reason across all four at once.

Neither piece is telling operators to buy anything specific. Both are making the same bet: the properties that treat data unification as infrastructure to build now, rather than an integration project to get to eventually, are the ones positioned to actually use whatever AI capability shows up next — while everyone else is still reconciling four dashboards to answer one question.