Cloudbeds' Josh Graham: AI Is the First Tech Wave That Favors the Independent Hotel

Cloudbeds’ Josh Graham argues AI is the first technology wave in decades that favors the independent hotel over the chain — because context beats scale, and independents hold real-time property knowledge that big brands typically lack.

Speaking from the HITEC 2026 show floor, Graham says the internet, cloud, and mobile eras all favored large chains; AI reverses that pattern. He calls the traditional property management system “an obsolete relic built for 1990s-era digitization” — he wanted to stage a mock funeral for the PMS at HITEC — and contrasts it with Cloudbeds, which he says was built from inception on a single unified data model rather than stitched together through acquisitions. His core warning: “scattered data becomes fatal” in the AI era. A spa integration that only posts charges isn’t enough; capturing a specific guest preference — a particular massage style, requested oils — lets AI trigger a relevant offer the moment availability opens. Cloudbeds’ AI product, Signals, pairs a “causal intelligence layer” tying rate, inventory, and marketing together with “Ask Signals,” a conversational interface with full platform visibility. Graham’s dividing line: probabilistic AI, which guesses, versus deterministic revenue systems, where guessing isn’t allowed.

Graham’s “scattered data becomes fatal” claim lands directly on the fault line exposed at a closed-door PMS roundtable the same week, where hotel CIOs and PMS vendors split 7-5 on whether the PMS stays central to the stack in five years, and the closest thing to consensus on data ownership was “whoever signs the contract owns the data” — no resolution for how records split across the multiple vendors touching one stay. Graham’s answer is architectural: unify the data model before layering AI on top. HotelKey’s co-founders offer the sharper counterpoint from ten days earlier: they’re openly skeptical of vendors relabeling rules-based automation as “AI” to capture budget, drawing the line at whether a system reasons and learns versus simply executes a fixed rule — close to the same test as Graham’s deterministic-versus-probabilistic distinction, from the other side of the podium.

For independent operators, Graham’s advice is concrete: audit the tech stack, eliminate data silos, consolidate vendors, build real AI fluency — rather than chase flashy point-solution agents. Whether independents actually hold the scale advantage he describes will show up in occupancy and RevPAR data long before it shows up in another vendor’s HITEC keynote.