Highline Hospitality Is Fixing Its Data Foundation Before Building In-House AI — the Same Sequencing Sage Hospitality Already Named

Highline Hospitality Partners, which operates 21 hotels and roughly $1.7 billion in assets under management, is betting that its in-house AI initiative can’t get built until its data problem gets solved first — so the company selected hospitality data platform Otelier, which serves more than 10,000 hotels worldwide, to centralize financial and operational data across its portfolio’s property-management and accounting systems. Otelier’s IntelliSight product gives Highline real-time, normalized data access through more than 100 pre-built dashboards spanning property-, department-, and transaction-level detail. HHP CEO Key Foster was explicit about the sequencing: “Otelier adds scale and stability to that foundation and will accelerate the AI capabilities we are developing” through the company’s in-house technology group, Highline Intelligence — data infrastructure first, proprietary AI second.

That sequencing echoes a framework Sage Hospitality’s CTO Matt Schwartz laid out a month earlier on Hospitality Daily’s AI series: a four-layer model — data, reporting, insights, action — where hotel data has to be consolidated and normalized before AI can produce useful analysis, reporting has to work before insights are trustworthy, and only then does autonomous action become viable. Schwartz’s explicit premise is that reliable action requires a trustworthy data foundation first, not a shortcut to it. Highline’s Otelier deal is the same maturity model showing up as a real vendor transaction rather than a stated philosophy: a mid-sized ownership group choosing to solve data normalization before building anything AI-branded on top of it.

For hotel groups still deciding where to spend their first AI dollar, both cases argue against the instinct to buy a visible, guest-facing AI feature first. The unglamorous data-normalization layer is what two separate operators, at different scales, both identified as the actual bottleneck — and skipping it doesn’t get skipped later, it just gets discovered.