AI for Hotels, Airlines, and TMCs That Actually Integrates

BCG and NYU published a joint report in March 2026 showing that fewer than 10% of hospitality companies qualify as "future built" for AI. Meanwhile, 37% of travelers already use LLMs embedded in OTA platforms to plan and book trips, and Sabre is launching an end-to-end agentic booking pipeline covering 420+ airlines and 2 million hotel properties this quarter. The gap between what travelers expect and what operators can deliver is widening, and closing it is not a software purchase. It is an integration, data, and compliance problem that cuts across your PMS, revenue management system, channel manager, CRM, and loyalty platform. That is where we work.

The Revenue Management Trust Problem

Hotels using AI-driven revenue management report a 17% revenue increase over non-adopters. AI-powered forecasting improves accuracy by roughly 20% versus legacy models, and dynamic pricing lifts ADR by 10 to 15%. Those are real numbers. The problem is that most properties never reach them, because revenue managers override the system. The trust gap between algorithmic pricing and human intuition is the last-mile failure of every RMS deployment we have seen. IDeaS G3, Duetto GameChanger, Atomize, and the newer entrants all produce defensible recommendations. But when the GM walks in and says occupancy "feels" soft for next Tuesday, the override happens. We build the explainability layer that earns trust: signal decomposition showing which inputs drove each rate recommendation, confidence intervals that tell revenue managers where to trust the model and where to apply judgment, and audit logs that satisfy the EU AI Act's August 2026 requirement to document why every price was set and how staff can review and override AI decisions.

Guest Data Is in Five Systems and None of Them Talk

A returning guest books through Booking.com. Their loyalty profile sits in your CRM. Room preferences are in Opera PMS. Their last spa booking is in a separate system. F&B history lives in your POS. The AI personalization vendor you just licensed cannot access any of it without custom integration work, and the integration cost is higher than the AI license itself. Mews raised $300 million in January 2026 at a $2.5 billion valuation to become the "hospitality operating system," and Duve raised $60 million in Series B for AI guest management. Both are building toward unified guest profiles, but both work best within their own ecosystems. If you run Oracle Opera for your flagship properties and Cloudbeds for your boutique portfolio, no single vendor unifies your guest view. We build the middleware: identity resolution across systems, preference aggregation from PMS/CRM/loyalty/POS, and the data pipeline that feeds your AI personalization with a complete guest picture rather than fragments.

Your AI Chatbot Is a Liability Until It Is Grounded

Air Canada deployed a customer service chatbot that hallucinated a bereavement fare refund policy. The policy did not exist. A customer relied on it, booked a flight, and filed a complaint. A Canadian tribunal ruled that Air Canada was liable for the chatbot's fabricated information and had to honor the non-existent policy. That is the precedent every hotel and airline is now operating under. Separately, CNBC reported in March 2026 that AI travel planners routinely recommend hotels that do not exist, attractions that closed years ago, and restaurants that were never real. Travelers have been directed to private residences described as "hidden gems." We build grounding and verification layers for hospitality AI: every property recommendation checked against live inventory databases, every policy response validated against your actual terms, every pricing quote confirmed at the point of commitment rather than the point of search. The 66% of guest inquiries that Conduit's benchmark shows can be handled automatically should be handled correctly.

Staffing Is Down 18% and AI Is Filling the Wrong Gaps

Sixty-five percent of hotels reported staffing shortages in 2025. Housekeeping is the hardest position to fill (38% of openings), followed by front desk (26%). Labor costs are up 11.2% year over year, a 25.6% increase compared to 2019. AI can help, but most deployments target the wrong problem. The Ritz-Carlton San Francisco implemented AI-synchronized housekeeping scheduling that cut room preparation time by 20%. That is the right application: augmenting constrained staff, not replacing them. Mobile check-in and digital keys reduce front desk workload by up to 40%. AI-powered food waste tracking cuts waste by roughly 50% within eight months. We help properties identify which operational bottlenecks actually respond to automation, build the integrations between scheduling systems and occupancy forecasts, and avoid the trap of buying AI tools that require more staff to manage than they save.

The Distribution Shift: NDC, GDS Surcharges, and AI Agents

Lufthansa Group is raising its GDS surcharge to $22 per ticket on Amadeus from May 2026, pushing agencies toward NDC. American Airlines has made 40% of its fares NDC-exclusive. But NDC has seven major versions in active use, and most mid-tier carriers have not built APIs worth connecting to. For TMCs, this creates a dual-channel reality: some content only available through NDC, some only through EDIFACT, pricing that differs between channels, and no single interface that handles both reliably. Layered on top: Google is building agentic booking into AI Mode search, Booking.com and Expedia are embedding AI trip planners, and Sabre's Red 360 platform uses an AI agentic layer with a Model Context Protocol. Eighty-five percent of GDS bookings already flow through APIs rather than human agents. The distribution landscape is fragmenting faster than any single vendor can unify. We build the integration layer that lets TMCs and hotel groups manage NDC and GDS content through consistent workflows, validate pricing across channels before commitment, and maintain control over inventory distribution as AI agents begin booking on travelers' behalf.

FTC, DOT, and EU AI Act: Three Regulators, One Pricing Page

The FTC's drip pricing rule took effect May 12, 2025: hotels must display total price upfront, including resort fees, destination fees, and service charges. Civil penalties reach $53,088 per violation. The DOT's airline fee transparency rule, implemented March 2026 after a court challenge, requires airlines to disclose baggage, seat assignment, and change fees during initial booking. The EU AI Act, enforceable from August 2026, classifies hotel dynamic pricing and guest profiling as high-risk when automated decision scope is broad enough, requiring price-setting logs, documented override processes, guest notification when AI profiling influences offers, and registration in the EU AI database. Fines reach 35 million euros or 7% of global turnover. These three frameworks interact. A hotel chain operating in the US and EU needs pricing display logic that satisfies FTC total-price rules, AI pricing documentation that satisfies EU AI Act audit requirements, and data handling practices that satisfy GDPR guest profiling constraints, all flowing through a revenue management system that was built before any of these rules existed. We instrument existing RMS platforms with the compliance layer: logging, explainability, and transparency features that satisfy regulators without requiring a system replacement.

Why Not Accenture, Deloitte, or Your Platform Vendor?

Accenture's minimum hotel engagements start at $500,000 or more. For a 50-property mid-market group trying to unify guest data across Opera and Cloudbeds, that is not a viable budget. Deloitte and McKinsey bring strategy and organizational transformation expertise, but when the problem is that your Duetto rate recommendations are not reaching your channel manager because the API mapping breaks on room type codes, you need an engineer who reads API documentation and understands revenue management workflows, not a strategy deck. Platform vendors build excellent products within their ecosystem. IDeaS optimizes pricing within IDeaS. Mews unifies operations within Mews. Neither has an incentive to build the cross-vendor integration that most real hotel tech stacks require, because most properties run three to five vendors that were never designed to work together. We are vendor-neutral, we work at the integration layer, and we charge for engineering hours rather than transformation programs.