Back
I am seeing an emerging theme as AI reshapes the hospitality industry. You can have the fanciest neural networks money can buy. The latest language models. The most advanced agentic platforms. None of it matters if you're feeding them yesterday's room inventory, last week's guest preferences, or booking and payment interfaces that AI Agents can’t navigate.
We're all guilty of chasing the shiny new toys. Everyone is rushing to showcase chatbots, predictive pricing analytics, Generative Engine Optimisation (GEO) and implementing Model Context Protocols (MCP). But here's what keeps me up at night: Are we building castles on quicksand? The conversation shouldn't be about which AI is the smartest. It should be about how we provision the cleanest, most reliable, real-time data and interfaces to feed those systems.
Picture this: A regular, loyal guest gets walked into a sold-out situation because the AI Agent assigning rooms during mobile self-check-in did not hold a room for this valuable customer.
That's not a glitch; that's a betrayal of trust. In fact, overselling should be a thing of the past! Or imagine your revenue system missing out because it doesn't know an airline crew's block of six rooms just opened due to a flight cancellation.
The now stranded passengers waiting for that same flight to turn around would've paid premium rates, but your system, and the downstream Booking Agents, never knew the rooms were available. Distressed inventory should also be a thing of the past!
The unglamorous truth about SORs and emerging AI protocols
I'll admit something: I have always been fascinated by the concept of Systems of Record (SORs) in the hospitality technology industry. My colleagues are probably tired of hearing me go on about it. But as we dive headfirst into AI, Agentic Commerce and Model Context Protocols, this concept becomes more crucial than ever.
What are Model Context Protocols (MCPs)? Model Context Protocols are emerging standards designed to help AI systems, especially Large Language Models (LLMs) and Autonomous Agents, interact more effectively with technology. Think of MCPs as universal adaptors or context orchestrators: they provide the business-specific context and structured interfaces that AI Agents need to understand and operate within complex tech stacks.
Think of the guest journey as a relay race. When someone's browsing hotels at 2 a.m., the search engine, website, or Online Travel Agency (OTA) holds the baton. They track every click, every saved property. Book through an OTA? They own that reservation until the guest walks through your door.
That's when your Property Management System (PMS) takes over. It becomes the single source of truth for everything: guest profiles, room assignments, special requests, rates, inventory, and folios. It's the conductor of the entire symphony. Meanwhile, your payment system handles the financial side, tokens, transactions, authorisations, and posting charges back to the folio.
Check-out triggers another handoff. Your Customer Relationship Management (CRM) and loyalty systems pick up the baton to nurture that relationship until next time. Each system needs to excel at its specific job. Asking your PMS to handle loyalty programs is like asking your sommelier to cook the steak. Sure, they work in the same restaurant, but they have different expertise. The AI agents need to know where to go, for what, and at the appropriate step in the journey.
AI Agents and the MCPs serving them need to know when the PMS is the SOR, which payment integration has authorisation for each step in the journey to remove the friction of re-entering payment information, and when to go to the loyalty system instead of the PMS for profile information.
The magic happens when they coordinate perfectly in terms of timing and execution.
Integration in the real world
Here's another truth: integration isn't something you do once and forget. It's like maintaining a garden; constant attention, regular updates, occasional major overhauls. The real challenge? Making sure the right system has the right data at exactly the right moment.
Your PMS talks to dozens of other systems. But communication isn't enough. These systems need true collaboration. Open APIs are just table stakes. What matters is vendors who commit to maintaining connections, respecting data standards, and genuinely caring about outcomes.
Most importantly, everyone needs to agree on who owns which piece of truth at what time. As we hand off responsibility to AI Agents and a future of agentic commerce, this becomes even more critical. MCPs and A2A protocols will provide the context for the AI Agents, but all systems integrated as part of the tech stack need to be aligned and agree on an SOR for each step or use case.
Let me share a real example. Our partnership with Canary Technologies looks simple on paper; guests message the hotel through an AI chat system. The systems exchange reservation data. Staff don't jump between apps. Neat and tidy, right?
Dig deeper, and it gets interesting. The AI pulls live data from the PMS and has access to underlying services whilst applying Canary's conversation smarts. It learns from every interaction. Need room availability? Got it. Upsell or check in without having to re-enter payment information? Totally aligned. Restaurant recommendations based on past preferences? No problem. Special anniversary request? Already handling it. No human needed for routine operations.
Why does it work? Because both companies went beyond plugging in APIs. We aligned on the technical details that actually matter: security protocols that protect guest data, governance frameworks for compliance, single sign-on that doesn't drive staff crazy, data standards that let information flow smoothly and crucially, token exchange to facilitate pre-authorisation at every step in the guest journey. We additionally embedded workflows in each other's platforms.
What makes Planet unique in this endeavour? We have PMS, payments, and retail platforms under one roof. This unified approach to SORs enables us to simplify further the complexity that hampers AI implementation in hospitality. As protocols like MCP and A2A become the standard, this will become even more apparent.
The future of AI in hospitality won't be built on algorithms alone. It needs clean data, clear system ownership, and genuine partnerships. Only then can we truly unlock the promise of agentic commerce, transforming the guest experience through the use of emerging technologies like Model-Centric Protocols or the next standard that evolves.
Stay in English