Every travel company today claims to be “AI-powered.” The buzzwords are everywhere: personalization, curation, recommendations, intelligent assistants.
But scratch beneath the surface, and you’ll find most of these AI tools do the same thing: they suggest.
Ask for a beach holiday in Greece? You’ll get a handful of destinations, maybe a few hotel options, perhaps even a sample itinerary. But when it’s time to actually book your trip, to press the button that turns all that digital chatter into a real reservation, you’re suddenly back in 2015. Copying details, opening new tabs, manually entering credit card info.
In short: AI in travel today is stuck at the advice stage.
It’s a smart concierge that can talk, but not act.
And that’s the gap holding back the next real wave of innovation in travel. Travelers don’t want endless recommendations, they want simplicity and convenience, they want technology that gets things done.
The Reality Check: AI Isn’t Booking Your Trip
The current generation of large language models (LLMs) like ChatGPT, Claude, or Gemini can write poems about Tuscany, compare airlines, or suggest “hidden gems” in Tokyo. But none of them can actually book anything.
Why? Because these systems aren’t connected to the transactional backbone of the travel industry. They don’t have a native bridge to the APIs that handle availability, pricing, payments, and confirmations. They exist in a layer of intelligence, but not in the layer of action.
This is why the travel industry today feels flooded with half-steps:
- Chatbots that “recommend” but can’t confirm.
- Virtual assistants that “curate” but can’t transact.
- AI that lists great experiences only to find out later that they are fully booked on the dates considered
- Tickets that have increased in price by the time they are being booked manually
- Apps that show “AI-driven inspiration,” then hand you off to a booking engine.
For all the AI hype, the process still breaks where it matters most: at the point of execution. Time is wasted on going back and forth between the “ideal personal itinerary” and the reality of fares, availability, and payments of reservations.
The Missing Link: MCP (Model Context Protocol)
Enter MCP — the Model Context Protocol — a technology quietly redefining what’s possible for AI in travel. MCP acts as the bridge between LLMs and the operational systems that actually make bookings happen.
Model Context Protocol (MCP) is an open protocol that allows you to provide custom tools to agentic LLMs (Large Language Models) like for example in Cursor’s Composer feature.
In plain terms: MCP allows AI agents to move from talking about actions to performing them.
It does this by giving AI systems structured, secure access to APIs, whether those are for flights, hotels, payments, or loyalty programs. Instead of merely suggesting, an AI powered by MCP can send the booking request, process the payment, and deliver the confirmation number, all while maintaining the user’s intent and preferences in context.
Think of it as turning a conversation into a transaction.
A traveler could say:
“Book me a direct flight to Lisbon on Friday morning early, a connecting train to Nazaré with a return Sunday evening landing in London before 11pm. I also need a hotel in Nazaré from Friday to Sunday,under £500. Favor my preferred providers.
And instead of only providing a list of suggestions, an AI agent integrated with MCP could:
- Query flight data from trusted sources and automatically find the best flight-train combinations
- Filter results based on the traveler’s preferences, loyalty scheme memberships and budget.
- Seek validation of the selected itinerary and proposed payment card.
- Makes booking and securely processes payment.
- Send back a confirmed integrated itinerary (“super PNR”)and digital receipt, all within the same conversation.
That’s not personalization. That’s automation with intent.
Data Quality: The Foundation of Real AI Travel
Of course, automation is only as good as the data feeding it.
One of the biggest challenges in travel today is data bias: the hidden influence of commissions, partnerships, and ranking algorithms. The “best” hotel or “recommended” flight is often a reflection of who’s paying more for visibility, not what’s actually best for the traveler.
To move from recommendation to real action, AI needs access to high-quality, unbiased data.
That means:
- Rich, structured datasets that accurately reflect real-time availability and pricing.
- Unbiased decision logic — free from commission-based distortions.
- Transparent ranking systems that prioritize user preferences over commercial interests.
Only when AI systems are trained and powered by clean, equitable data can they truly act in the traveler’s best interest. Otherwise, automation just amplifies the same old biases, only faster.
The End of “Inspiration-Only” Travel Tech
For years, the travel industry has obsessed over inspiration: slick images, curated recommendations, and AI-written itineraries. It’s time to admit that inspiration is no longer enough.
The next evolution of travel technology isn’t about showing people where they could go, it’s about taking them there.
That means:
- LLMs that connect to real-world systems through protocols like MCP.
- A bridge between the outdated travel sector booking infrastructure and the AI world
- Data integrity that drives fair, transparent recommendations.
When all three converge, travelers will move from endless browsing to seamless booking. And the brands that enable that shift, the ones that automate action instead of generating noise, will define the future of the travel industry.
From AI Advice to AI Execution: Junction’s Vision
At Junction Connect, an AI native travel tech platform, we’re building exactly that future, where AI doesn’t just talk about travel, it makes it happen.
Our approach is simple:
- Use MCP to connect intelligent agents directly to booking APIs.
- Leverage clean, unbiased data to drive transparent choices.
Conclusion: The Future Belongs to the Doers
AI in travel is evolving. But real innovation will come from systems that act, that connect, transact, and deliver outcomes.
The industry’s next wave isn’t about generating ideas; it’s about eliminating friction. The winners will be those who stop talking about personalization and start automating execution. In doing so they will bring real value to the traveller.
Travelers don’t need more inspiration. They need technology that gets things done.
And with MCP at the core, that future is finally within reach.
Bio
Goran Burazer is Chief AI & Innovation Officer at Junction, building AI-driven travel infrastructure that unifies global content via a single API and enables agentic booking workflows using the Model Context Protocol. Since joining in 2015—later serving as CTO—his focus has been re-architecting travel’s data plumbing and real-time operations. Writing about applied AI, orchestration, and the future of frictionless multimodal travel.
