From AI Experimentation to AI Adoption: How Travel Companies Can Get More Value from Generative AI
Most travel companies do not begin their AI journey with a carefully planned strategy. Adoption tends to happen much more organically.
Someone in the business starts experimenting with a free version of ChatGPT. Another employee hears good things about Claude and tries that instead. Someone else experiments with Gemini. Gradually, AI starts appearing across the organisation, but different people are using different tools, for different purposes and with very different levels of confidence.
In our work with travel, tourism and hospitality organisations, we have found this to be a common pattern. Initial experimentation is valuable because it creates awareness and allows people to experience the potential of generative AI for themselves. The difficulty comes when the organisation needs to decide what happens next.
Should the company standardise around one AI platform? Which platform should it choose? Is it worth investing in paid licences for the whole team? How should those accounts be set up? What should employees actually be using AI for? When should the business move beyond individual prompting into reusable workflows, automation and more structured implementation?
These questions matter because using AI occasionally and adopting AI as an organisational capability are two very different things.
01
The AI maturity journey is often accidental
There is a recognisable maturity journey that many organisations follow with generative AI, but it is rarely planned from the beginning.
It typically progresses from initial awareness and individual experimentation towards paid adoption, coordinated team use, reusable workflows and eventually more sophisticated implementation and automation. Some organisations move through those stages relatively quickly. Others remain stuck at one stage for a surprisingly long time.
The earliest phase is usually straightforward. Employees hear about generative AI, create free accounts and begin trying it on tasks that feel safe and familiar. Writing is a natural starting point, followed by summarising documents, carrying out basic research, brainstorming ideas and improving emails.
Over time, several people within the business may be doing this independently. At that point the organisation can legitimately say its employees are using AI, but there may still be little consistency in how they are using it, which tools they have chosen or whether the business is receiving meaningful value from that activity.
This is the point at which experimentation needs to start becoming adoption.
“Using AI occasionally and adopting AI as an organisational capability are two very different things.”
02
Why businesses get stuck using free AI tools
Free versions of generative AI tools are important because they remove the financial barrier to experimentation. For somebody who has never used ChatGPT, Claude or Gemini, they provide an easy way to understand what these systems can do.
The problem arises when the free version becomes the company’s long-term AI strategy.
Moving to paid AI involves a genuine business decision. Purchasing software for an organisation is different from an individual paying for a personal subscription. Once multiple employees require licences, the company is making a recurring monthly commitment that may continue for years.
It is therefore reasonable for leaders to want confidence that they are choosing the right platform before committing. Unfortunately, deciding which platform to choose is becoming harder rather than easier.
In my own experience, I can open ChatGPT, Claude or Gemini and regularly find that something has changed since my previous visit. This is particularly noticeable with ChatGPT and Claude. The interface may have changed, a feature may have been renamed, a new capability may have appeared, or something that previously worked in one way may now work differently.
The pace of change is extraordinary. New models, capabilities and products appear constantly, and the volume of announcements, comparisons and conflicting opinions makes it increasingly difficult for a normal business user to know where to begin.
For someone running a destination management company, tour operator, hotel business or tourism organisation, keeping track of every development could become a job in itself. Their priority is running the business.
The result is often decision paralysis. Companies continue using free accounts because they are waiting for enough certainty to make a platform decision. Yet remaining on limited free tiers can also prevent them from properly evaluating the capabilities that become relevant once AI is used more seriously across a business.
03
Social media is not a technology procurement strategy
The amount of information surrounding generative AI creates another challenge: businesses need to decide whose advice they trust.
Social media is full of confident claims that one platform is dramatically better than another. Some of that content is excellent, but recommendations may also be influenced by sponsorships, affiliate relationships, personal preferences or use cases that bear little resemblance to the requirements of a travel business.
We regularly encounter people who have formed very strong views about particular AI platforms despite having relatively little first-hand experience of comparing them properly.
Claude is a good example. Its reputation has become so strong in certain online communities that people who have barely used the major AI platforms themselves sometimes arrive with the assumption that Claude must automatically be the best choice.
It may well be the right choice for them. But that conclusion should follow an assessment of the organisation’s requirements rather than precede it.
A company-wide software decision deserves more due diligence than choosing whichever platform currently receives the strongest recommendations on LinkedIn, YouTube or another social network.
04
There is no universally best AI platform
Businesses would be better served by moving away from asking, “Which AI tool is best?”
The more relevant question is: Which platform is best suited to our organisation, our employees, our existing systems and the work we want AI to support?
Claude may be the right choice for one organisation while ChatGPT makes more sense for another. An organisation heavily invested in Google Workspace may have good reasons to consider Gemini, while a Microsoft-centric business will have a different set of factors to assess.
The decision should take account of the work employees perform, the systems already used by the business, available integrations, privacy and security requirements, collaboration and administration features, the capabilities different teams genuinely need, ease of adoption and the overall cost of deployment.
This is one reason AI strategy and training increasingly overlap. Choosing a platform without understanding how the organisation intends to use it makes meaningful due diligence much harder.
05
Paying for AI does not automatically mean you are getting more value from it
Once a company finally selects a platform and purchases licences, another common problem appears.
Employees often continue using the paid platform in almost exactly the same way they used the free version. They draft emails, summarise documents, rewrite copy and conduct occasional research. These are all legitimate uses of generative AI, but they represent only part of what modern paid platforms can support.
Paid tools increasingly provide ways to retain useful context, work with larger amounts of company information, create shared workspaces, analyse files and data, conduct deeper research, develop reusable instructions and connect AI with other business applications.
Without training and some organisational structure, employees may never discover many of these capabilities, let alone understand how they could apply them to their own work.
This is why buying AI licences is not the same thing as creating AI adoption. The organisation may have upgraded its technology while leaving its working practices largely unchanged.
06
Good AI training should change the way people look at their work
The most important outcome from practical AI training is not simply teaching people how to write better prompts. It is helping them start looking at their own work differently.
An employee might initially use AI to draft a supplier email. With more experience, they may notice that they write a very similar type of supplier email several times each week. That leads to a more interesting question: should this remain a one-off prompt, or could the relevant context and instructions be captured so the process becomes reusable?
The same thinking can be applied across a travel business. Someone who repeatedly researches hotels may recognise that the same evaluation criteria could be used every time. An operations employee who manually compares information across several documents may begin asking whether AI could perform an initial review and flag discrepancies. A colleague who regularly transfers information from one place into another may identify an opportunity for a more structured workflow.
The important capability is not memorising a long list of things AI can supposedly do for travel businesses. It is developing the judgement to recognise where AI can remove friction from the work the organisation already performs.
This is also why generic platform demonstrations have limited value on their own. Teams understand AI much more quickly when examples reflect the realities of their own industry and responsibilities.
For a DMC, that may involve enquiries, proposals, itinerary development, destination research, guest communication, supplier relationships, reservations and operations. Another type of travel or hospitality organisation will have different priorities.
07
AI adoption needs organisational structure
As individual experimentation grows, organisations need to decide which parts of their AI setup should become shared company assets.
There is usually value in creating consistency around areas such as brand and writing guidance, company terminology, verification standards, responsible AI use and recurring processes used by several employees.
At the same time, not everything should be centralised. An employee responsible for supplier relationships may need a very different AI workspace from somebody working in marketing or finance. Individuals should have enough freedom to identify applications relevant to their own roles and experiment with better ways of working.
The aim is to create enough structure to improve consistency and reduce unnecessary duplication without creating so much control that employees stop experimenting.
The strongest organisations tend to develop both: a shared foundation and sufficient individual capability for people to improve their own workflows.
08
Why we recommend appointing an internal AI champion
Because generative AI changes so quickly, organisations also need somebody who maintains momentum after the initial training or implementation work has finished.
We increasingly recommend identifying an internal AI champion.
This does not need to be a developer, technologist or AI specialist. In many companies, the right person is simply the employee who is naturally most curious about the technology. They experiment with new capabilities, follow important developments, identify repetitive work, share discoveries with colleagues and keep asking whether there is a better way to perform a particular task.
That person provides valuable continuity. Given the speed at which ChatGPT, Claude, Gemini and other platforms are evolving, a business cannot realistically assume that one training programme will keep everybody current indefinitely.
Where somebody naturally takes on this role, they should be encouraged. Where nobody does, we recommend formally nominating an AI champion and giving them the space to develop that capability.
09
Training is important, but sometimes training alone is not enough
Another pattern has emerged from our work with travel and hospitality businesses.
A company approaches us because it needs AI training, but the discovery process reveals that there are also pressing business problems it wants AI to solve.
The organisation might have an administrative bottleneck consuming a disproportionate amount of employee time. Information may be scattered across several systems. A recurring process might involve repetitive checking or manually transferring information. Valuable company knowledge may sit in inboxes and individual employees’ heads rather than somewhere the wider business can use it.
Training is clearly part of the solution because employees need the skills and judgement to use AI properly. However, there is a practical limit to what can reasonably be expected immediately afterwards.
If a team has previously done very little beyond basic prompting, asking employees to complete a training programme and then immediately design robust workflows and automations is a significant leap. They may understand the opportunities without yet having the practical experience required to build everything themselves.
10
Closing the gap between training and implementation
There are two models we increasingly see working well.
The first is a combined programme in which strategy, consultancy, training and implementation happen alongside one another. Employees build their AI capability while we work with the organisation to identify and develop practical use cases around real business problems.
We are currently applying versions of this approach with two organisations in the travel and hospitality sectors. In both cases, the objective is not simply to train the team. It is to help solve genuine business challenges while developing the organisation’s internal AI capability at the same time.
The second model is sequential. Some organisations prefer to train their team first and then move directly into strategy, consultancy and implementation work afterwards.
We have found this to be another effective approach because it closes the gap between learning and doing. Employees complete the training with a much better understanding of what AI can support, and the organisation then moves immediately into identifying, prioritising and implementing the most relevant use cases rather than allowing the momentum from the training to dissipate.
Both approaches can accelerate adoption because employees are not left with a theoretical understanding of what they might build at some point in the future. They begin seeing how the principles apply to their own processes, data and business challenges.
This can also bring the organisation closer to measurable business value much earlier in its AI journey.
11
Different organisations need different starting points
There is no single AI adoption programme that makes sense for every travel business.
Some organisations come to us having barely used generative AI. Their first requirement is understanding what the technology can and cannot do, learning how to use it safely and building enough confidence to begin experimenting.
Others already have employees using several free AI platforms but lack a coordinated approach. They may need help comparing tools, choosing an appropriate company platform and deciding how it should be introduced.
Another business may already have purchased paid licences but find that most employees are still using only basic capabilities. Training can then focus more heavily on getting value from the existing investment, introducing reusable context and workflows, and identifying opportunities across individual roles.
Further along the maturity journey, priorities begin shifting towards implementation, integrations and automation.
The intervention should therefore begin with the organisation’s actual level of AI maturity rather than forcing every business through the same predetermined course.
12
Curated Greece: moving beyond general AI use
Our work with Curated Greece, a boutique destination management company, provides a good example of an organisation that had already moved beyond the earliest stages of adoption.
Curated Greece had invested in paid Claude access, and AI usage was already widespread within the team. A pre-training survey found that 87% of respondents were using AI for work at least weekly, so this was not a business that needed to be persuaded to start using generative AI.
The opportunity was to help employees get more value from the platform they already had.
The programme focused on stronger briefing and prompting, better judgement around AI outputs, more effective use of context, and moving from one-off conversations towards more repeatable AI use within individual roles. Between the sessions, participants applied the training to genuine work and began identifying recurring activities that could potentially develop into more structured workflows.
A company whose employees had never used generative AI would require a different starting point. That is precisely why assessing AI maturity matters before designing the training.
13
What does successful AI adoption actually look like?
AI maturity should not be judged simply by how many licences a company owns or how many automations it has built.
A more relevant measure is whether AI has become a practical capability within the organisation.
Employees should understand where AI can help and where it cannot. They should know how to provide enough context to generate strong output and when that output needs to be verified. The organisation should have made considered decisions about the platforms it uses, how they are configured and which standards or knowledge should be shared.
People should also be identifying recurring activities where reusable workflows could remove unnecessary effort. More advanced teams may begin connecting AI with other systems and automating selected processes, while maintaining appropriate human review where mistakes could have significant consequences.
Most importantly, the business should eventually be able to connect its AI activity with outcomes that actually matter, whether that means time saved, increased capacity, better consistency, improved service, fewer missed details, increased revenue or another measurable objective.
That is a more meaningful definition of AI adoption than simply being able to say that everyone has an AI licence.
14
Do not wait for the AI market to settle
The speed of development understandably makes some organisations reluctant to make decisions. It can feel sensible to wait until the market becomes clearer before choosing a platform or investing heavily in AI.
The difficulty is that the market is unlikely to become stable enough for businesses to stop thinking about it.
Models will continue changing. Interfaces will change. Features will be renamed, replaced and introduced. Prices and integrations will evolve. A provider leading in one capability may be overtaken elsewhere.
For most businesses, the answer is therefore not to predict which AI company will eventually win. It is to develop skills and working practices that remain valuable even when the underlying technology changes.
Knowing how to brief AI properly, provide relevant context, evaluate outputs, design repeatable processes, manage risk and identify worthwhile business applications are transferable capabilities. They remain relevant whether an organisation uses Claude, ChatGPT, Gemini, Copilot or whatever comes next.
The platform matters, but the organisation’s ability to use AI intelligently matters for much longer.
15
Where should a travel company start with AI?
The right starting point depends on where the organisation is today.
A business with almost no experience of generative AI should concentrate first on understanding the technology, establishing responsible practices and building practical confidence. Where employees are already experimenting independently, the priority may be creating some direction, assessing the available platforms and deciding whether the company should standardise around a paid tool.
If licences have already been purchased, the organisation should examine whether employees are genuinely using the capabilities it is paying for and whether enough structure exists around AI use.
Once adoption becomes more established, attention can shift towards reusable workflows, shared AI assets and selected opportunities for automation.
Where the company also has significant business problems that AI could help solve now, training can be combined with, or immediately followed by, strategy, consultancy and implementation so the organisation develops its people and its practical AI capability together.
The goal is not to use AI everywhere. It is to understand where AI can make the business better and how to implement it properly when it can.
16
AI training and adoption for travel, tourism and hospitality organisations
Digital Dialog has more than a decade of experience working with travel, tourism and hospitality organisations. We bring that sector knowledge into our AI work from the outset, including an understanding of the commercial pressures, customer journeys, operational complexity and workflows that are specific to these industries.
That means our clients do not need to teach us how a travel or tourism business works before we can start discussing where AI may be relevant. We can focus more quickly on the organisation itself: its current level of AI maturity, its people, its technology, its priorities and the business problems it wants to solve.
We support organisations at every stage of AI maturity. Some clients are starting from the beginning and need help understanding the technology and building confidence across their teams. Others already have employees using generative AI and need help choosing tools, improving adoption or getting more value from platforms they are already paying for. More advanced organisations may need support developing workflows, solving specific operational challenges and implementing AI more deeply across the business.
Depending on the requirement, our work can combine practical AI training with strategy, consultancy and hands-on implementation.
Whether your team has never used generative AI or is already paying for Claude, ChatGPT or another platform, Digital Dialog can help you identify the right next stage and turn AI into a practical business capability.
See our AI case studies, including our work with Curated Greece for examples of how our approach changes depending on an organisation’s starting point.
Manu Kastia
Manu Kastia is Founder and AI consultant at Digital Dialog, an AI consultancy specialising in tourism, travel and hospitality. With over 15 years of experience, Manu's expertise encompasses AI strategy, training, and advisory services for the sector. He has successfully worked with major brands including Switzerland Tourism, British Airways, Eurostar, Tourism Ireland, and Marketing Manchester. Manu's passion for making AI practical and accessible has positioned him as a sought-after speaker at industry events and a trusted consultant for organisations across tourism, travel, and hospitality. He helps businesses navigate AI decisions through strategic advisory, hands-on training, and comprehensive AI literacy resources. Manu has played a pivotal role in advancing AI knowledge through training sessions and strategy consulting, empowering professionals to harness AI for genuine business outcomes. His extensive sector background and practical approach make him a trusted advisor for those looking to navigate AI opportunities with confidence.