What Happens When AI Becomes the Travel Agent?
Imagine planning a three-day European city break and typing:
“Find me a destination with step-free public transport, a hotel with a roll-in shower, museums with quieter visiting periods and no more than one major transfer per day.”
For many travellers, that kind of request would once have meant opening dozens of tabs, searching for individual pieces of information, contacting hotels and attractions, checking transport routes and trying to work out whether the different parts of the journey would actually connect.
Increasingly, AI can help bring some of that process together.
Google's AI travel tools can already create personalised itineraries using information from Search, Maps, hotels, flights, reviews and websites, while newer features are moving towards helping users turn plans into bookings.
That does not mean AI has replaced the travel agent. But it does point towards a significant change in how people may discover where to go, where to stay and what to do.
And for inclusive travel, that raises an important question:
Can an AI system answer a traveller's specific needs accurately, or is it simply making its best guess from fragmented information?
Travel discovery is changing
Travel discovery has traditionally involved searching, browsing and comparing.
A traveller might search for accessible hotels, open several booking platforms, read reviews, visit destination websites and then contact individual businesses to clarify what "accessible" actually means.
AI changes the starting point.
Instead of translating a need into a series of keywords, a traveller can describe the experience they want in natural language and ask a system to narrow down the options.
Google describes this shift as moving from keyword matching towards understanding intent. Its travel tools can take a detailed request, bring together information from multiple sources and help users compare options based on factors such as price, amenities and location.
That could be particularly valuable for travellers who have more complex requirements.
For someone who needs step-free transport, a particular type of bathroom, quieter environments and manageable distances, reducing hundreds of options to a small shortlist could remove hours of research.
But there is a catch.
Narrowing the field is only useful when the information used to narrow it is accurate.
An AI system that identifies five genuinely suitable hotels could make travel planning dramatically easier. An AI system that incorrectly identifies five hotels as suitable could create even more uncertainty.
That means the question is no longer only whether a destination appears in search results.
It may increasingly be whether that destination is selected for the shortlist an AI system generates.
What information does AI actually use?
There is no single, perfect database containing everything an AI system needs to know about a destination.
Travel information exists across supplier websites, booking platforms, business listings, destination-management websites, reviews, maps, images, public transport information and other digital sources.
Some information may be structured. Some may sit within paragraphs of website copy. Some may only appear in a review. Some may be out of date.
And different sources can contradict each other.
One hotel might describe itself as accessible. A booking platform might provide no further detail. A recent traveller might mention a step at the entrance that neither source records.
This creates an important distinction between information that is:
available online, discoverable by a person, interpretable by a machine and reliable enough to support a recommendation.
Google's own guidance for travel businesses highlights the importance of detailed, structured content, noting that AI can only surface information it can access and interpret.
Structured information is not particularly complicated in principle. Instead of simply saying:
Accessible room
a business could provide clearly labelled information such as:
Room type: accessible twin room
Entrance: step-free
Bathroom: roll-in shower
Door width: 920mm
Turning space: 1,500mm
Accessible parking: available 30m from entrance
Last verified: June 2026
That makes information easier for systems to identify and compare.
But structure is not the same as truth.
A machine-readable claim can still be incomplete, inaccurate or out of date.
The quality of future travel discovery will therefore depend on more than whether organisations add structured data. It will depend on whether the information is specific, complete, accurate, verified and regularly updated.
The problem with "accessible"
This becomes particularly obvious when looking at the word "accessible".
"Accessible" is often treated as an answer.
For many travellers, it is only the beginning of a list of questions.
Does it mean step-free access? Is the whole route step-free? Is there an accessible toilet? What are the dimensions? Is there enough space beside the bed? Is the shower level access? Is there accessible parking? How far is it from the entrance?
And suitability can vary between travellers.
A hotel might meet the needs of one wheelchair user but not another. A museum might have step-free access but be overwhelming for someone with sensory sensitivities. An attraction might provide assistance, but only at certain times.
Research into the online travel experiences of disabled travellers has highlighted the importance of reliable information in reducing uncertainty when planning and evaluating hospitality and tourism services.
A useful way to think about accessibility information is through three layers:
Presence: Is the feature available?
Specificity: What exactly is available?
Suitability: Will it meet this particular traveller's needs?
Consider the difference between:
"Accessible bathroom."
and:
"Level-access shower with a shower seat, grab rails and a 1,500mm turning circle."
The second gives a traveller much more to work with.
But even that may not answer every question. Photographs, diagrams, independent assessments, staff knowledge and lived-experience reviews can provide context that a simple data field cannot.
This is why inclusive travel information should not be reduced to a tick-box exercise.
Whose needs are represented in travel data?
The issue goes beyond accessibility.
Travel data tends to reflect what organisations have chosen to describe, and the categories that platforms have chosen to provide.
That can leave gaps.
A neurodivergent traveller might want information about noise, lighting, queues and predictable layouts.
An older traveller might need realistic walking distances, seating and rest points.
A Muslim traveller might want to know about prayer facilities, halal food or nearby services.
A family might need adjoining rooms, changing facilities or step-free routes suitable for a buggy.
Someone with allergies might need detailed information about ingredients and cross-contamination.
A traveller managing fatigue might need to understand how much walking an itinerary actually involves.
These needs do not always fit neatly into the categories currently used by travel platforms.
And that matters because the traveller who is easiest to represent in a database is not necessarily the traveller with the fewest needs. They may simply be the traveller for whom the industry has created the clearest data fields.
AI could potentially help expand how travellers describe what they need. But the systems still need enough reliable information to respond.
Could AI make inclusive travel easier to find?
There is a genuine opportunity here.
AI could help a traveller turn a personal requirement into relevant search criteria. It could ask follow-up questions, compare destinations, summarise lengthy access information and bring together different parts of a journey.
It could potentially help someone identify a hotel, nearby attraction and transport route that work together rather than evaluating each one separately.
That could reduce the amount of research required before booking.
More importantly, it could reduce some of the uncertainty that makes travel planning difficult for people whose needs are poorly represented by standard booking filters.
The most valuable AI travel experience may not even be one that produces an instant, confident answer.
It could be one that says:
"These three options appear to meet your requirements. This information has been verified. This detail is based on traveller reviews. This part could not be confirmed, so you may want to contact the provider."
That distinction matters.
For inclusive travel, making uncertainty visible may be more valuable than pretending it does not exist.
Or will AI reproduce the gaps we already have?
AI will not create inclusive travel from nothing.
It will amplify the quality, visibility and limitations of the information already present in the travel ecosystem.
If accessibility information is vague, an AI system may have little basis for making a reliable recommendation.
If information is outdated, it could repeat an outdated claim.
If a smaller business has an excellent accessible experience but very little digital information, it could be overlooked in favour of a larger business with more structured and searchable content.
That creates a potential form of digital visibility bias.
The businesses most visible to AI systems may not necessarily be the businesses offering the best experiences. They may simply be the ones with the most complete, connected and machine-readable information.
This is why organisations should be cautious about treating AI as a neutral filter.
An algorithmic shortlist can look objective while reflecting the inequalities and gaps in the data used to create it.
There is also a risk of collapsing different access requirements into broad categories.
A system might interpret "wheelchair accessible" as a complete answer when, in reality, accessibility depends on the entire journey: transport, entrances, rooms, toilets, attractions, routes and staff support.
The challenge is therefore not simply to make more information available.
It is to make better information available.
The new question for destinations: Are you AI-readable?
For tourism organisations, hotels, attractions and destinations, this creates a new question.
Not:
"Have we published accessibility information?"
But:
"Can the right traveller discover, understand and trust that information when they are making a decision?"
That means looking at the information ecosystem as a whole.
Is accessibility information specific rather than generic?
Is it consistent across the website, booking platforms and destination listings?
Does it cover the journey rather than a single property?
Are different needs represented?
Is the information regularly updated?
Are limitations clearly stated?
Can travellers see when information was last verified?
Is important content available as accessible, human-readable text rather than being hidden in images or inaccessible documents?
And where structured data is being used, does it actually match what the traveller sees on the page?
These are not simply technical questions.
They are questions about the quality of the traveller experience.
Research into accessible tourism has already highlighted the importance of technology and online information in helping travellers with disabilities plan and navigate tourism experiences. As AI becomes another layer through which that information is discovered, the stakes could become higher.
Inclusion becomes part of digital visibility
This is where the conversation moves beyond accessibility and into destination strategy.
If AI increasingly influences which destinations, hotels, attractions and experiences enter a traveller's consideration set, then accurate representation becomes increasingly important.
Inclusive information could affect whether a traveller discovers a business at all.
That does not mean organisations should start producing content simply to "rank in AI".
The goal should be the opposite.
Make the real experience clear, specific and trustworthy, and make sure that information is available in formats that both people and digital systems can understand.
That is good communication.
It is good traveller experience.
And it could also be good digital infrastructure.
The future isn't about designing for AI
The future of travel discovery should not be about designing experiences for machines.
It should be about designing better experiences for people, then making those experiences accurately visible.
AI can help travellers ask more nuanced questions.
It can potentially connect information that has historically been fragmented.
It could make inclusive travel easier to discover than it has been before.
But it cannot compensate for information that does not exist, cannot distinguish fact from outdated claims without reliable signals, and cannot know what a traveller needs if those needs are not represented in the information available to it.
So perhaps the more useful question is not:
"Can AI plan my holiday?"
It is:
"When I tell AI what I need, will it understand enough about the destination to help me make a good decision?"
For the travel industry, that creates a practical responsibility.
Inclusive experiences need to be more than possible.
They need to be visible, specific, understandable and trustworthy before the traveller arrives.
Because if AI becomes part of the journey between "I want to travel" and "I've chosen where to go," the information destinations provide today could determine who gets to discover them tomorrow.
Help shape the future of inclusive travel
Is your organisation thinking about how to make its inclusive offer easier for travellers to discover, understand and trust?
The Inclusive Travel Forum works with destinations, tourism organisations, hotels, attractions and travel businesses through training, trade guides, destination visits, thought leadership and tailored partnerships.
Explore how we can work together
You can also join the Inclusive Travel Forum to connect with travel professionals, access practical resources and be part of the conversations shaping a more inclusive future for travel.