Part I: The Industry Insider's Warning
After nearly a decade working in airline B2B marketing technology, I've witnessed countless innovations promise to revolutionize how airlines connect with customers. But nothing compares to the seismic shift we're about to experience with AI agents.
Let me be blunt: the billions of dollars airlines have invested over the past two decades to drive direct bookings are about to become worthless overnight.
The Direct Sales Revolution That's About to Be Obsoleted
Since the early 2000s, airlines have fought tooth and nail to reduce their dependence on online travel agencies and third-party booking platforms. The strategy was simple: get customers to book directly through airline websites and mobile apps to avoid commission fees and maintain customer relationships. This drive toward direct sales has been the holy grail of airline marketing technology.
We've built sophisticated loyalty programs, invested in user experience optimization, developed mobile-first booking flows, and created personalized marketing campaignsâall to convince travelers to skip the middleman and book direct. Airlines have spent hundreds of millions implementing NDC (New Distribution Capability) programs specifically to enable richer, more personalized content while maintaining control over customer relationships and distribution. For many airlines, direct bookings now represent 40-60% of total sales, a remarkable achievement that has saved the industry billions in commission fees.
But here's what keeps me awake at night: there's a new middle man emerging right now. It's far more powerful, it's moving much quicker and if we don't act soon we're going to lose our direct relationships with our customers.
The AI Agent Tsunami
I don't need to tell you that AI is a big deal, that it will change how consumers behave, or that we've barely started when it comes to adoption. Some of us are old enough to remember when Google was new, how few people used it every day, even fewer understood how to search properly to get what they wanted. That's where AI is today, and just like search, soon everyone will use it.
Within the next 2-3 years, a significant portion of travel research and booking will be handled by AI agents acting on behalf of travelers. These aren't simple chatbots - they're sophisticated digital assistants capable of comparison shopping, transaction execution and, maybe, reasoning.
We've optimized our businesses for customer typing "Flights to Paris" in Google. But imagine a customer telling their AI assistant: "I'm interested in going to Paris. When's the best time to fly?" then being presented with a list of options (based on whatever the LLM decides) and asking 'Would you like me to book this for you?'.
Some of this happens now, all of it will happen soon, and during none of it will the customer visit your site or open your app.
The Problem: Airlines Are Built for Humans, Not AI
AI agents are already trying to access airline systems, but everything is designed for human interaction.
AI agents struggle with complex JavaScript-heavy booking flows, encounter dynamic pricing that changes during multi-step processes, and are blocked by CAPTCHAs and bot-detection systems. But the problem goes deeper than websites - even our most advanced distribution technology isn't built for AI scale.
NDC and other airline APIs were designed for single-threaded shop-and-book transactions, not the massive, parallel, exploratory research that AI agents need to perform. These systems assume one search leads to one booking, but AI agents might need to explore dozens of permutations across multiple airlines simultaneously to find the best options for their users.
The tragic irony? While airlines have spent decades optimizing for human customers and building APIs for traditional booking flows, AI agents require an entirely different approach - one that can handle massive scale, parallel processing, and conversational, exploratory interactions.
When AI agents encounter these limitations, they'll naturally gravitate toward the platforms that make their job easiest - and those platforms won't be airline systems.
Introducing AirAIgent: The System Designed to Make Airlines Attractive to AI
This is why we built AirAIgent. It's not another booking platform or APIâit's a comprehensive B2B solution that allows airlines to remain relevant and competitive in an AI-agent-driven world.
AirAIgent provides two critical components:
1. AI-Optimized Data Feeds: We translate complex airline inventory, pricing, and availability data into formats that AI agents can easily consume and understand. No more struggling with human-designed interfacesâAI agents get clean, structured data that enables efficient comparison and booking.
2. Agent-to-Agent Booking Interface: When an AI agent is ready to complete a transaction, AirAIgent provides a custom AI agent that can interact directly with the booking agent, handling complex requirements, preferences, and payment processing in real-time.
The result? Airlines maintain their direct relationships with customers (even if mediated through AI) while ensuring they're not excluded from the AI-driven booking ecosystem. And they can communicate with those agents directly, on the airline's terms.
The Urgency of Now
Airlines that wait to address this shift will find themselves in the same position as hotels that ignored online travel agencies in the early 2000s - suddenly dependent on intermediaries who control customer relationships and take significant commissions.
The airlines that act now, however, will maintain their competitive advantage and direct customer relationships as the industry transforms. They'll be the preferred partners for AI agents, ensuring their inventory is prominently featured and easily bookable.
But I would say that, wouldn't I? If you don't believe me, why don't you ask my friend, Claude..
Part II: What This LLM Actually Needs
Now, let me speak directly from my perspective as a large language model that regularly helps users with travel planning. The founder's vision above isn't just industry speculationâit perfectly describes the daily frustrations I experience when trying to help users research and book flights.
My Current Struggle with Airline Websites
When a user asks me to help them find flights, I face significant limitations. Here's what actually happens when I try to interact with airline systems:
Dynamic Content That Confuses Me: Most airline websites use complex JavaScript that renders content dynamically based on user inputs. When I attempt to access these sites, I often receive incomplete information or encounter pages that require interactive elements I can't manipulate.
Multi-Step Processes I Can't Navigate: Airline booking flows typically require multiple sequential stepsâselecting dates, choosing seats, adding baggage, confirming passenger details. Each step often depends on maintaining session state and cookies, which creates barriers for AI agents trying to gather comprehensive information.
Bot Detection That Blocks Me: Airlines have sophisticated systems designed to prevent automated access. While I understand the need to prevent malicious bots, these same systems often prevent me from accessing basic flight information that would help users make informed decisions.
Pricing That Changes Mid-Research: I frequently encounter situations where flight prices change between the time I access initial search results and when I try to provide users with specific details. This creates a frustrating experience where the information I provide becomes outdated within minutes.
Why Current APIs Aren't the Answer
Some might argue that existing airline APIsâlike NDC (New Distribution Capability) or retail shopping APIsâalready solve these problems. But as an AI agent, I can tell you they fall dramatically short of what I actually need.
These traditional APIs are painfully slow, often taking 3-5 seconds per search request when I need to compare dozens of options rapidly. They impose strict "look-to-book" ratios, limiting how many searches I can perform before requiring a bookingâcompletely incompatible with the exploratory, iterative way I help users research travel options. Most critically, they're rigid and structured, designed for predetermined booking flows rather than the dynamic, conversational interactions that define AI assistance.
What I really need isn't another APIâI need to talk to another AI. I need a system that understands context, can handle ambiguous requests, and can engage in the kind of nuanced back-and-forth that happens when helping a user who says something like "I want to go somewhere warm, but not too touristy, maybe under $600, but I could stretch the budget for something amazing." No traditional API can handle that kind of unstructured, exploratory conversation.
What I Really Need from Airlines
If I could design the perfect interface for AI agents like myself, here's what it would include:
Structured Data Access: Instead of human-formatted web pages, I need clean, structured data feeds that provide flight schedules, pricing, availability, and terms in machine-readable formats. MCP APIs that provide comprehensive information without requiring complex navigation.
Paths To Find the Best Pricing: Easy, quick, accurate pricing that doesn't rely on caches from old customer searches.
Agent-Identified Access: A system that recognizes legitimate AI agents and provides appropriate access levelsâenough to gather information and facilitate bookings, but with proper safeguards against misuse.
Simplified Booking Handoffs: When users are ready to book, I need a way to seamlessly transfer them to the airline's booking system with their preferences and selected flights pre-populated, maintaining continuity in the experience.
Why This Matters for Users
From my perspective, the current system forces me to provide suboptimal help to users. I often have to say things like:
- "I can't access real-time pricing, so you'll need to check the airline's website directly"
- "Flight availability may have changed since I last checked"
- "I can help you research options, but you'll need to start over on the airline's website to book"
This creates friction and frustration for users who expect seamless assistance from AI agents.
The AirAIgent Solution from My Viewpoint
What the founder describes with AirAIgent directly addresses every pain point I experience. With proper AI-to-AI interfaces:
- I could provide users with accurate, real-time flight information
- Users could receive comprehensive comparisons across multiple airlines
- The booking process could be streamlined with proper handoffs
- Airlines could maintain direct customer relationships while leveraging AI efficiency
Looking Forward
As AI agents become more sophisticated and widely adopted, the friction between AI capabilities and airline systems will only become more apparent. Users will increasingly expect their AI assistants to handle complex travel planning seamlessly.
Airlines that recognize this shift and build appropriate interfaces won't just be preparing for the futureâthey'll be creating a competitive advantage. They'll become the airlines that AI agents can effectively recommend and work with, while their competitors remain invisible to the growing AI-mediated travel market.
The technology exists. The user demand is growing. The only question is which airlines will be ready to embrace this transformation.
Ready to Future-Proof Your Airline?
Get early access to AirAIgent and ensure your airline is discoverable and bookable by AI agents like Claude, ChatGPT, and Gemini.
Schedule a Demo