Being involved with an AI Incubator (Neural River) gives me access to bright young AI minds constantly monitoring new AI tools and developments.

Google’s new Jarvis AI tool for Chrome initially did not excite me as, in essence, it’s just an AI Driver “RPA Tool”. This is where a computer program can mimic the “key stokes” of a human being to complete a booking process.

These technologies have been in use for over a decade. In collaboration with the company ‘Traverse,’ I have automated various travel processes, such as booking verification and accounting reconciliation for low-cost carrier bookings, by extracting data from confirmation emails and matching it with back-office booking databases.

Once the team began experimenting with the tools, I quickly realised what a remarkable asset it was. Unlike RPA tools, which often fail whenever a webpage’s format changes, the new Jarvis tool utilises live website imagery and integrates information from multiple sources, achieving a nearly 98% success rate.

Airline XML fee wars.

Recently, I highlighted an ironic shift in the travel industry: Ryanair, by charging zero API fees, has unexpectedly become online travel agencies (OTAs) best friends after years of battling each other. Meanwhile, On the Beach (OTB) has intensified its war with EasyJet by offering a ‘Self-Book and Save’ option that allows a family of four to save £96. Through this scheme, post-booking OTB provides customers with a virtual credit card and a deep link, guiding them to book their flights directly on EasyJet’s site. Though somewhat cumbersome, many customers accept it as a way of saving such a large amount.

Google’s Jarvis tool could automate this process into a two-click procedure, removing 90% of the effort and potentially dooming API fees.

Comparison Engines.

Comparison engines such as Skyscanner and Icelolly direct traffic to supplier websites for booking completion. While this approach often maximises click-out revenue relative to acquisition costs, it significantly limits their ability to foster repeat bookings, as they lack visibility into customer data post-booking.

Leveraging Jarvis to complete bookings could soon encourage these platforms to adopt a model like Amazon’s, positioning themselves as the retailer. This would enable them to capitalise on cash flows while providing customers with a smoother, faster booking experience.

The pressure to implement this approach will intensify as AI-driven search engines, like www.Perplexity.ai (Try it), gain prominence, with customers increasingly valuing engines that provide direct ‘answers’ rather than a list of links. Naturally, these models will prefer connecting users to streamlined booking pathways that validate the best value. This suggests a comparison engine that allows customers to complete bookings directly on-site.

Homeworkers Co-pilots.

My AI-voice business, Neural Voice, is already working on Homeworker “Co-Pilot” tools to answer the phone and gather customer requirements out of hours or when they are busy with other tasks.

Research indicates that homeworkers spend 60% of their time on administrative tasks like booking flights online or entering bookings into back-office systems. By automating data loading via Chrome, we could reduce this workload by 75%, allowing homeworkers to focus more on providing a personalised booking experience for customers.

Although the same tools will make booking easier for customers, they will not replace the core function of travel homeworkers, which is providing expert human advice and guidance.

As I’ve often noted, it’s astonishing how slowly the travel industry has embraced the AI revolution. If you haven’t yet recruited a young AI-focused team, you may already be at risk of becoming a “Dinosaur” ready for extinction as the travel landscape rapidly evolves.

If artificial intelligence or AI scares you, you may want to stop reading now, as the implication of Elon Musk’s “Neuralink” project place is firmly in the “Frankenstein” horror genre.

Much discussion has been focused on the rapid development of AI and the road to artificial “General” intelligence, generally defined as when computers using AI become more intelligent than the entire human race combined. Many observers rightly worry about what role is left for the human race when AI superintelligence makes humans look dumb and irrelevant.  

The human brain is an extraordinary organic computer, with memories manifesting as electrical impulses between neurons. Our intelligence stems from consciousness, encompassing self-awareness and the ability to plan based on past experiences and future predictions. Individual memories, largely unshared collectively, drive our daily actions. Tragically, when we die, this unique intelligence and personal history typically vanish, which is massively wasteful.

AI systems, in contrast, learn collectively and preserve knowledge indefinitely. Envision a toddler mastering walking and instantly sharing this skill with all peers worldwide. This exemplifies how AI supercomputers function, explaining the exponential growth of AI technology.

Interestingly, humans are relatively slow communicators. Our brains Humans communicate, on average, at 1 bit per minute over a day, mainly via the spoken word or keystrokes.

Our brains process ideas and turn them into words to communicate them to other humans or computers via keystrokes. This forces a distillation of an idea to speak it, which is good. However, the effective transfer rate depends on how other humans interpret these words and can visualise the person’s intent. Not surprisingly, there is a lot of leakage and inefficiency here.

Humans can only communicate at 1% of the rate of computers, making us massively inefficient in their eyes. From their perspective, they must wait for an eternity to respond to their prompts for information or a response.

Through his well-funded Neuralink business, USA entrepreneur Elon Musk is focused on building a faster connection between computers and the human brain.

The company is developing implantable brain-computer interfaces (BCIs) to help those with paralysis and neurological conditions in the near term and enhancing human-AI interaction in the long term.

Neuralink’s core technology consists of ultra-thin probes inserted into the brain to detect neural signals and custom-designed electronics and software processes used to interpret brain signals. The company’s first device, the N1 Link, is a 1024-channel implant designed to be invisible once inserted and capable of wireless data transmission.

In recent interviews, Musk was clear that although the early focus was helping disabled quadriplegics and blind people overcome neural damage, this was just because the “Rewards” were significant enough to counterbalance the risk associated with this new procedure.

However, as the number of transplants grows and the risk diminishes, he expects these tools to become freely available, creating humans with superpowers.

For example, suppose Neural links can stimulate the brain to replicate regular visual sight. Why not also give the user the ability to have infrared or radar, as the computer providing the service supports these technologies?

Neuralink’s long-term vision is to enable direct brain-computer communication, potentially accelerating human-machine interaction tenfold by bypassing traditional input methods. This neural interface could theoretically provide instant access to vast online knowledge repositories, potentially revolutionising human creativity and cognitive capabilities.

This technology could theoretically enable the transfer of human memories and decision-making processes to digital systems, preserving individual cognition beyond death and potentially creating collective intelligence. Such advancements raise profound questions about the nature of consciousness, identity, and the future of humans as a race.

Whether such “Frankenstein” technologies will benefit humanity is yet to be seen, but the complete lack of regulation over AI Ethics should concern everyone.

AI development is currently dominated by companies with the largest “Compute” resources. However, the next frontier in advancing artificial general intelligence (AGI) will likely centre on access to high-quality training data. Elon Musk’s strategic investments in X (Twitter) and Tesla’s Optimus robot project potentially position him to acquire vast amounts of real-world data, which could prove crucial in realising his AI ambitions.

Where Musk is “Jekyll or Hyde,” his vast wealth allows him to steer humans’ future in the direction of his choice with little ethical oversight. Personally, I admire the guy, but just thinking through the logical conclusions to this technology scares the hell out of me.

Welcome to the future!

Based on Sam’s 2-hour interview with Lex Fridman

Interestingly, Sam regards ChatGPT 4 as just an announcement to the human race of what is coming and believes that history will define the start of the AI Revolution as when it allows thousands of scientific breakthroughs that benefit humans from cures to significant diseases or understanding whether and where alien life may exist in the universe.

So basically, we have not seen anything yet, and Open AI is attempting to release iteratively so that they don’t scare us to death with one massive lead forward we are not pre-prepared for.

The ultimate goal is AGI (Artificial General Intelligence), which, for newbies to the topic, is defined as when artificial intelligence encompasses the ability to understand, learn, and apply knowledge in a way that is not limited to a specific task or domain, or discipline”. In other words, machines are capable of independent thought.

The race between Elon Musk, Amazon, Google, Microsoft and Open AI, to name a few, is clearly on, and the winner will yield massive power.

Altman stresses that there must be clear board frameworks to ensure that no one man can yield the full power of AGI, but openly admits that the recent failure of the Open AI board, misguided or not, to successfully remove him as CEO, shows that his own business is not yet structured correctly.

Personally, having listened to several of Alman’s interviews, I am glad he is leading the charge towards AGI. He is incredibly intelligent, reasonably humble, and appears to have a clear moral compass.

But hey, I liked Boris at first, so what do I know!

Surprisingly, Altman is against adding advertising revenue into Open AI’s income stream because it will add “bias” to the models and potentially distort the route to the “right answer” for the customer. He prefers that customers pay higher fees for ad-free, unbiased models, but he accepts that competition may force his hand.

The biggest restrictor of AI in the short term is “Computing Power,” and Altman flagged that AI needs to start getting smarter at allocating and charging for computing resources between simple questions and answers to complex mathematical formulas. Currently, Open AI charges a flat fee per response, ignoring the processing cost of creating the response.

Altman talks about “Compute” being the currency of the future and believes we need to power these computers via nuclear fission or fusion. He flagged that it is the “theoretical”  fear of Nuclear meltdowns like Chornobyl that has put humans off nuclear energy, when in reality, the death toll from the pollution from coal power stations has been much higher and that the future needs to be a clean nuclear if we are to avoid further Global warming.

Sadly, Sam Altman also believes it’s likely that the controversy over AGI is likely to trigger his assassination by a gun-toting American reactionary, preventing him from seeing the benefits that AI can deliver to the human race.

Unfortunately, I think of all his prophecies, this is the most likely to be right.

Voice-activated search is set to explode, with companies like Neural Voice allowing users to talk to the knowledge of the Internet in a conversational way.

The business has created a wide range of AI characters to demonstrate its capabilities, ranging from lawyers to travel agents who advise on the best holiday destinations.

For example, travel companies can email AI Emily, a virtual travel agent, to previous bookers to find out where they want to go on their “Next Holiday.” Customers then have a conversation with AI Emily about their holiday needs. To achieve this, the system transcribes what they say into words and injects a question into a “Large Language Model” (LLM) to receive a response that turns back into human-sounding speech, all within 0.84 seconds.

Hence, AI Emily’s responses are as fast as a normal human’s but with unlimited knowledge about the world’s top holiday destinations. Emily can guide customers through a choice of destination, right down to individual hotels, while giving customers a full review of all nearby restaurants and bars.

Once Emily has done her job, the customers are emailed back a summarised transcript of their call. The customer’s requirements are also injected into the holiday company’s “Customer Response Marketing” (CRM) system to ensure all future advertising messages are based on their “Next Holiday” requirement and not random holiday offers aimed at getting them to visit the company’s website.

However, although this technology is ideal for most generations, it is likely to be rejected out of hand by the “Betamax” generation of 18–34-year-olds who hate voice interaction. Research shows that 70% will not even pick up the phone when their parents call. Even when they do listen to the message, they prefer to reply by text.

This generation was brought up in a communication world dominated by shorthand text and WhatsApp messaging. They prefer to search by typing into Google and regard parents using voice-activated search with disdain.

I have renamed this “Gen Z” age group the  “Betamax Generation” because, like the video format that battled for supremacy in the mid-80s with the alternative VHS format, the writing is already on the wall for the end of their toys.

Children who cannot yet read and write can use Voice activation to control TVs, open doors, and turn on lights, so there is no way they will revert to typing to search the Internet.

Similarly, the over 35’s brought up in phone conversations have no issue talking to a phone to drive a search query and will be the early adopters of the new Voice-based platforms.

This leaves the “Betamax Generation” out in the cold, with technology rapidly evolving to make their text-based environments redundant.

Interestingly, all business SEO strategies must evolve to allow for the much longer and more complex search terms used by AI search engines.

AI Engines like www.Perplexity.ai have already gathered 10 million plus users. They are likely to rapidly eat into Google’s 95% dominance of internet search as its results summarise everything you need to know whilst providing links to the originating articles so that users can validate their conclusions.

The future of search is voice-activated and AI-delivered. You must have the lights off if you’re not reading that writing on the wall.

As predicted, Ryanair has now made stage 2 of its API integrations for OTAs available. Instead of presenting a higher “Bundled” price, these integrations allow OTAs to display the lead “Basic” Ryanair price and charge upgrades for extras such as seat selection, speedy boarding, and hold luggage.

This instantly makes Ryanair prices via OTA’s appear cheaper, as they are now on a like-for-like basis with Easyjet and Jet2.com, increasing their market share with a minimal impact on revenue, as most customers will continue to choose these upgrades when offered.

Interestingly, on the Ryanair site, these options are bundled so that if you want speedy boarding, you must also select a seat; within OTAs, these are broken down as separate charges.

However, as usual, the highly commercial Ryanair has priced “hold luggage plus seat” at the same price via OTAs but offers a lower price on its site for combining the popular “Speeding Boarding” options with seat selection. This option is required if you’re on a short break and want to carry 10kg of hand luggage, which is why most Speedy boarding queues are longer than standard boarding.

I am sure that other Ryanair OTA partners, like On the Beach and Tui, will quickly follow suit in upgrading to this new API connection as soon as their technology backlog allows.

Ryanair’s offering OTAs price parity will pressure low-cost carrier competitors to follow suit. For example, Easyjet’s £6.00 per passenger per flight sector equates to £26.50 for the average 2.2-passenger booking, with Jet2.com yielding their charge based on route demand and partner volume.

However, the true cost of providing API connectivity is in pence rather than pounds. These fees were introduced to compensate for lost ancillary revenue and provide in-house tour operations a price advantage. However, they look increasingly unsustainable in a market with excess flight capacity.

Overall flight capacity from the UK is forecast to remain at 98% of 2019 levels in 2024, primarily due to weaker demand for business travel post-COVID-19’s mass adoption of video conferencing as a feasible alternative to face-to-face meetings and a reduction in demand for city breaks.

Low-cost carriers have instead shifted large amounts of capacity onto leisure routes. Exactly how much capacity has been added is hard to estimate, but with bonded ATOL carryings having increased by 5.32M or 20% since 2019 to 31.6m in 2024, the overall increase will likely be more than 10m seats.

This extra capacity is already dragging down average load factors and forcing low-cost carriers to either drop flight prices or offer substantial discounts off package holidays where they can dump excess seats opaquely.

So, it looks like a tough summer ahead for low-cost carriers but a bumper year for OTAs and in-house low-cost tour operators. I have always impolitely described OTAs as “parasites living off the misery of others” with distressed outbound seats from one airline combinable with other airlines’ seats, often providing market-leading flight prices.

The OTA’s ongoing threat, however, is that the airlines use their tour operations as the route to dispose of excess seats, giving them a significant price advantage and driving massive growth.

That is why, in my opinion, the latest market is still likely to be painted “Orange”, with Easyjet Holidays growing dramatically as it provides an alternative seat filler for its parent airline.

Recent Phocus Wire research showed that the top 4 global travel brands spent an astounding $16.8 billion in 2023 on sales and marketing, up 20% from 2022. An estimated 70% of this, or $11.8, was handed over to Google in the form of search advertising.

The Monopolies and Mergers Commission would swiftly initiate an investigation if a merger proposal arose among major low-cost carriers or if an acquisition attempt was made on an Online Travel Agency (OTA) such as On the Beach or Love Holidays. However, they ignore entirely Google’s dominant market position, which holds 93.5% of the UK’s online search market share.

How can one business be allowed such a dominant position, with virtually total control over how UK customers access online holidays? Particularly when Google does not even pay a fair level of UK tax, siphoning off its UK profits to a lower-tax Southern Ireland office.

However, a quick review of the table below highlights how Airbnb is bucking this trend.

Airbnb is proud that it spends relatively little on Google marketing and drives 90% of traffic on a non-paid basis directly to its website.

It helps that they have such a clearly defined product proposition and loyal repeat booking customer base, but they also focus heavily on creative social media and PR stunts, such as collaborating with Martell to turn a Malibu mansion into the “Malibu Barbie Dream House” listing on Airbnb. This alone generated millions of site visits.

Airbnb is also investing heavily in influencer marketing, providing tools to its owners to carry out their own marketing, and driving app usage by ensuring that all communication with the host goes through this channel.

In the UK, On the Beach (OTB) still spends 51% of its revenues on marketing, but unlike direct competitors Love Holidays, it is spending increasing amounts on “Above the Line” on TV and Radio to create brand awareness and drive more direct traffic. The benefits of this brand building may take years to develop, and Love Holidays has rapidly caught OTB in terms of turnover and exceeding its profit performance in 2022.

Love Holidays’ recent accounts to October 31st, 2022, showed a turnover of £ 1.35 billion, just behind OTB £1.44 billion but with an impressive EBITDA of £36.3m. Clearly, their excellent technology is driving conversion and customer retention levels that allow them to pay high Google acquisition costs and scale rapidly, making Google an effective traffic source.

If you’re not one of the major players, reliance on Google is probably a deferred death sentence waiting to happen. So, what alternatives are emerging?

1. Perplexity Search. Playing with this tool by visiting www.perplexity.ai/ will demonstrate how quickly it delivers “answers” and should move you away from Google search. Although the tool has 10 million uses across Asia, India, and the USA and is valued at $520m, it’s relatively new to the UK, having only launched in August 2022.

Why would the customer continue to spend 24 hours researching and clicking on 50 Google links when they can get an answer in an instant from Perplexity or ChatGPT? As my recent blog outlined, Open AI’s CEO Sam Altman is opposed to advertising, which he believes will add “bias” to results. He prefers to offer a premium subscription model, and this is indeed the model Perplexity operates.

However, history has shown us that although new online tools initially shun advertising during their user growth period, it always creeps in. In the meantime, travel SEO plays like little-known www.cheekytrip.com/ , is ranking well and harvesting lots of free traffic.

2. Database Marketing. Most users now ignore millions of emails, with opening rates and click-throughs for travel emails falling rapidly. However, new AI technology offers a potential solution for re-invigorating this approach.

Customers can now be emailed a link to an AI Travel Agent such as Emily www.neural-voice.ai/characters/emily/, who can have a conversation informed by their previous travels to establish their requirements for this year’s trip. The conversation is automatically transcribed, summarised, and posted into the business CRM to contextualise future advertising or for the lead to be followed up with online booking links or a call from a human agent to close the sale.

3. Influencers. If you are not an Instagram user, you will not comprehend the power that influencers can yield over the “Dream Stage” of a holiday. This medium is compelling for tourist boards and upmarket hotels and is increasingly becoming a proven traffic generator. Recruiting your own team of influencers or working with influence aggregators like Trending Travel www.trendingtravel.co.uk/ should be considered.

4. Homeworkers. This is the ultimate low-cost route to market, with localised marketing via sports clubs and businesses supported by targeted social media activity controlled by the homeworker who understands their local market dynamics. However, companies like Travel Solutions Network www.travelsolutionsnetwork.co.uk/ are now offering business access to “Virtual Call centres staffed by commission-only homeworkers, to offer high converting “Human agents” to complement their online booking path. Research has shown that showing that a business is “Contactable” in the case issues emerge increases “online conversion”, making the phone bookings 100% incremental, even after booking route substitution.

As Love Holidays has proven, Google search as a customer acquisition route can still work if you have great technology that converts these leads. However, for the rest of us, it’s probably time to focus on exploring some of the new routes to market listed.

As a child, I often kept score at the Darts Board for the adults, which considerably sharpened my mental arithmetic and may have led to my early career as an accountant. However, the advent of hand-held calculators and spreadsheets made these skills semi-redundant so that I would now avoid the darts chalkboard like the plague.

Does this make me less intelligent?

Today, we have numerous examples of where technology has killed the use of elements of our brain, ranging from day-to-day Navigation (Who doesn’t use their phone) to the ability to follow a set of instructions without a supporting YouTube “How to video”.

Again, does this make us less intelligent or more productive because we have co-pilots that make us more efficient?

The advent of the internet has had a massively positive impact on our efficiency and day-to-day lives, but it has also made us lazier.

Why bother walking around high-street shops when you can pop online, read reviews, choose items, have them delivered, and send them back if they don’t suit you?

In the same way, why not use previous online baskets to repeat supermarket orders delivered to your doorstep?

However, the critical question is, what are people doing with the extra time created?

The good news is that many use it to get fitter by riding bikes, going on long hikes, or enjoying team sports; however, just as many use it to spend more time on social media, gaming, or watching on-demand TV often making them unfit “couch potatoes”.

In a recent interview, Sam Altman, CEO of Open AI, said that ChatGPT should be viewed as announcing what is coming rather than the start of the AI Revolution.  He believes the start will occur within the next ten years, when AI enables scientists to produce 1,000s of new cures for human diseases, considerably lengthening life expectancy.

Science has long proven that Humans use only a fraction of their brains, which gives me hope that AI will be a co-pilot that will make many humans much more intelligent and able to make significant scientific breakthroughs that will benefit us all.

The race for AGI or “Artificial General Intelligence” concerns me more. AGI is when artificial intelligence encompasses the ability to understand, learn, and apply knowledge in a way that is not limited to a specific task, domain, or discipline”. In other words, machines are capable of independent thought.

Sorry, but why would a computer that is smarter than a human need a human as a Co-pilot?

If AI and robotics replace relatively dull human labour in factories and call centres, it might not be such a bad thing, but AGI will replace lawyers, accountants and doctors, the realm of our most intelligent humans.

The significant danger is that 80% of humans will have less work stimulating their brains and multiple tools to make their daily lives easier. Therefore, their intelligence is likely to decline and not increase.

Conversely, the 20% of humans who increase their intelligence will likely become a “Super Wealthy Elite”, creating even bigger gaps between rich and poor increases, causing social unrest.

How we manage the implementation of AI and AGI will ultimately determine the future of the human race, and the debate needs to start now.

What do you think?

Alphabet, Google’s owner, will never disappear from our lives as it has the power and wealth to “buy” the next internet sensation or solution, as it did with YouTube. But are we about to see a seismic shift away from Google’s 95% dominance over internet search?

For a generation, Google has held a monopoly over travel searches. However, this dominance comes at the cost of a relatively poor customer experience. Research reveals a staggering truth: the average customer is forced to visit 48 travel sites and invest a whopping 26 hours in research before finally booking a holiday. Imagine the frustration level out there!

Google’s search services are monetised through a “Cost per Click” (CPC) or CPA model. The more clicks it generates, the higher its profits. But have you ever noticed that the longer the search you type into Google, the less likely you are to find the result you want?

Google encourages shorter search terms and returns pages of links for customers to click on to research further to find the answers they need to take the “Action” they require, e.g. booking a holiday. These results are further distorted by “paid bias”, with advertisers bidding to be placed in the most prominent positions.

Commercialisation has become so bad that 28% of the UK population actively uses Adblockers, which, for Google searches, may indicate a desire to get the right answer quicker.

Large Language models (LLMs) are exploding in popularity as they appear to offer a much faster route to the “right” answer.

LLMs are trained on the entire knowledge of the internet and are much better positioned to interpret long and complex enquiries by focusing on the “intent” of the keywords used to reach a better understanding of the customer’s question.

They also allow a query to be “built” by considering what the user has said previously in the same query stream, allowing customers to refine their requirements based on the results returned at each stage.

This has allowed Neural Voice, one of my AI start-ups, to create a “Voice” overlay where customers can talk to an AI character, in a conversational format to discuss options for where they want to go on holiday next year (Dream Stage). Check it out; the sophistication of the tech, even at this early stage, will amaze you https://www.neural-voice.ai/

Links to Emily, our AI agent, can then be emailed out to previous bookers to qualify their requirements for their next holiday, with a summary and transcript automatically passed to the customer’s CRM to influence further marketing or for follow-up by a human agent.

LLMs and voice interaction promise to offer a much better customer experience, allowing them to find the information they need for “action” much faster. This could quickly erode Google search volumes while providing a new route to market for the travel business. However, the model is radically different, and the LLM’s commercial model will quickly evolve.

Here are a few things for travel businesses to start thinking about:

Commercial model.

Like most people, I am currently happy to pay my $20 monthly licence fee to use ChatGPT. However, this subscription revenue will quickly get eroded by competition from other LLMs, and inevitably, these models will seek to add transactional revenues.

However, adding bias to direct customers to partners paying higher advertising fees, in the same way, Google does, would be diametrically opposed to the ethos of providing the best answers without bias and would create a clunky endpoint.

I personally think LLMs will follow the Amazon model.

Amazon is, in effect, a massive “retail” search engine, where the customer chooses from a huge range of suppliers’ products but transacts with and pays Amazon. Travel comparison sites like Trivago have also evolved to transact this way as it increased conversions by removing friction created when customers clicked out to partner sites.

So, if LLM delivers an “end-to-end” booking service, what are the other knock-on impacts?

Will SEO and quality of content matter in the longer term?

Many travel businesses are rushing to use AI technology to create new SEO content, to drive them up the SEO rankings. However, although some LLMs integrate with Google content, most models are based on online data-rich sites like Wikipedia rather than individual website pages, making them inaccessible via traditional SEO optimisation.  Getting LLMs to prefer your travel products will be a key battle, but it’s very unclear how this will be done yet.

Secondly, if LLMs are transactional, they will likely use their content database and differentiate between suppliers based on price rather than content quality.

Creating and protecting travel “brands”.

The large airline-owned tour operations of Easyjet and Jet2 holidays have natural protection in that customers travel on branded flights and are serviced by branded ground staff delivering in-resort services.

However, OTAs like Love Holidays and On the Beach’s main role is to provide a wide range of products and a simple booking process. What is their added value if the LLMs replace this booking process?

Obviously, there is an opportunity for these brands to use LLM technology to become the next Amazon of travel, but this does seem like a big ask and although not all travel traffic will flow through LLMs, it does seem like these players could be long-term losers.

Google Search, like the Dinosaurs of yesteryear, may quickly become extinct, but what replaces it and how the travel industry integrates into this new search model is far from clear.

However, the more dependent your business is on Google search traffic today the faster you need to act.

Alphabet, Google’s owner, will never disappear from our lives as it has the power and wealth to “buy” the next internet sensation or solution, as it did with YouTube. But are we about to see a seismic shift away from Google’s 95% dominance over internet search?

For a generation, Google has held a monopoly over travel searches. However, this dominance comes at the cost of a relatively poor customer experience. Research reveals a staggering truth: the average customer is forced to visit 48 travel sites and invest a whopping 26 hours in research before finally booking a holiday. Imagine the frustration level out there!

Google’s search services are monetised through a “Cost per Click” (CPC) or CPA model. The more clicks it generates, the higher its profits. But have you ever noticed that the longer the search you type into Google, the less likely you are to find the result you want?

Google encourages shorter search terms and returns pages of links for customers to click on to research further to find the answers they need to take the “Action” they require, e.g. booking a holiday. These results are further distorted by “paid bias”, with advertisers bidding to be placed in the most prominent positions.

Commercialisation has become so bad that 28% of the UK population actively uses Adblockers, which, for Google searches, may indicate a desire to get the right answer quicker.

Large Language models (LLMs) are exploding in popularity as they appear to offer a much faster route to the “right” answer.

LLMs are trained on the entire knowledge of the internet and are much better positioned to interpret long and complex enquiries by focusing on the “intent” of the keywords used to reach a better understanding of the customer’s question.

They also allow a query to be “built” by considering what the user has said previously in the same query stream, allowing customers to refine their requirements based on the results returned at each stage.

This has allowed Neural Voice, one of my AI start-ups, to create a “Voice” overlay where customers can talk to an AI character, in a conversational format to discuss options for where they want to go on holiday next year (Dream Stage). Check it out; the sophistication of the tech, even at this early stage, will amaze you https://www.neural-voice.ai/

Links to Emily, our AI agent, can then be emailed out to previous bookers to qualify their requirements for their next holiday, with a summary and transcript automatically passed to the customer’s CRM to influence further marketing or for follow-up by a human agent.

LLMs and voice interaction promise to offer a much better customer experience, allowing them to find the information they need for “action” much faster. This could quickly erode Google search volumes while providing a new route to market for the travel business. However, the model is radically different, and the LLM’s commercial model will quickly evolve.

Here are a few things for travel businesses to start thinking about:

Commercial model.

Like most people, I am currently happy to pay my $20 monthly licence fee to use ChatGPT. However, this subscription revenue will quickly get eroded by competition from other LLMs, and inevitably, these models will seek to add transactional revenues.

However, adding bias to direct customers to partners paying higher advertising fees, in the same way, Google does, would be diametrically opposed to the ethos of providing the best answers without bias and would create a clunky endpoint.

I personally think LLMs will follow the Amazon model.

Amazon is, in effect, a massive “retail” search engine, where the customer chooses from a huge range of suppliers’ products but transacts with and pays Amazon. Travel comparison sites like Trivago have also evolved to transact this way as it increased conversions by removing friction created when customers clicked out to partner sites.

So, if LLM delivers an “end-to-end” booking service, what are the other knock-on impacts?

Will SEO and quality of content matter in the longer term?

Many travel businesses are rushing to use AI technology to create new SEO content, to drive them up the SEO rankings. However, although some LLMs integrate with Google content, most models are based on online data-rich sites like Wikipedia rather than individual website pages, making them inaccessible via traditional SEO optimisation.  Getting LLMs to prefer your travel products will be a key battle, but it’s very unclear how this will be done yet.

Secondly, if LLMs are transactional, they will likely use their content database and differentiate between suppliers based on price rather than content quality.

Creating and protecting travel “brands”.

The large airline-owned tour operations of Easyjet and Jet2 holidays have natural protection in that customers travel on branded flights and are serviced by branded ground staff delivering in-resort services.

However, OTAs like Love Holidays and On the Beach’s main role is to provide a wide range of products and a simple booking process. What is their added value if the LLMs replace this booking process?

Obviously, there is an opportunity for these brands to use LLM technology to become the next Amazon of travel, but this does seem like a big ask and although not all travel traffic will flow through LLMs, it does seem like these players could be long-term losers.

Google Search, like the Dinosaurs of yesteryear, may quickly become extinct, but what replaces it and how the travel industry integrates into this new search model is far from clear.

However, the more dependent your business is on Google search traffic today the faster you need to act.

Ryanair’s recent shift in perspective, from being anti-OTA, a stance they often referred to as battling ‘Pirates,’ to now viewing OTAs as strategic distribution partners, has pleasantly surprised many in the travel industry.

Love Holidays deserves recognition for being the trailblazer in bringing Ryanair to the table. However, the initial deal, with variable booking fees of between £30-50 per couple, seemed steep and posed a potential price disadvantage against other OTAs, such as On the Beach (OTB), if they continued to ‘screen scrape’ and avoid these fees. Strategically, Love needed the deal to prove to potential new investors that they had long-term access to cheap flight seats departing the UK, which is the lifeblood of their business.

On the Beach then entered negotiations and would appear to have improved the terms of the deal.

On the Beach’s site boldly tells customers that their Ryanair flight prices now include pre-booked seats with Ryanair and, at the same time, having monitored Love’s Ryanair prices since January, in March, Ryanair’s extra fees on the Love site suddenly changed to a fixed £32.00 return per couple or £8.00 per flight sector to match those on the On the Beach site.

Connecting these two pieces of information, it is now clear that Ryanair is not charging an API fee but is making it compulsory for people booking holidays to pre-book their seats at a fixed cost of £8.00 per flight, which is probably the average pre-booked seat price across their network.

This compulsory addition seems odd for a business that prides itself on breaking down the price to create the cheapest possible lead price and then charging separately for extras, as it makes their prices appear less competitive. However, suppose you had not planned for a massive increase in third-party trade distribution. In that case, it is unlikely that your third-party APIs would be geared up for complexities such as allowing customers to choose and pre-book individual seats.

I would expect Ryanair to resolve this over the next few months, allowing them to return to supplying stripped-down prices with no booking fees that will dominate the OTA’s search results. This will boost volumes further and put pressure on EasyJet and Jet2.com to consider the size of their API fees.

EasyJet currently charges around £6.00 per sector or £24 per couple, while Jet2.com takes a much more tactical approach, charging zero fees on weaker selling routes.

As I said when the deal was announced, “On the Beach are no mugs,” but their Ryanair deal looks better by the day because it not only delivers a guaranteed supply of cheap seats but also puts pressure on other low-cost airlines to become more OTA-friendly.

The deal will also remove a massive element of back-office cost because OTB will no longer have to hide from Ryanair using virtual credit cards, roving IP addresses, and doing lots of manual work to notify customers of Ryanair schedule changes and cancellations. I’d estimate this cost saving will run into the millions and could lead to a substantial profit boost.

The “Ryanair War on OTA’s” is over, and the only question is, why did it take so long?  

Nobody outside of Ryanair’s inner circle will ever know, but ironically, the change of heart seems to have stemmed from Ryanair’s success in the Irish courts in declaring the “screen scraping” of their flights illegal. Although technically unenforceable in other source markets, this emboldened them to massively tighten up all their tools to stop OTA sales in November 2023, including the highly contentious extra facial recognition steps for “illegal” OTA customers.

This suddenly closed all their OTA distribution and forced Ryanair to announce to the city that this was impacting their load factors by 1-2% and reducing average flight prices. It does not take a genius to guess that the city might have asked, “What on earth are you doing?”

We will never know the answer, but the attitude reversal has been dramatic. Now that all parties see the benefits of working together, it’s hard to see why this would ever reverse for packaging OTAs’, but I am far less convinced Ryanair will embrace flight-focused OTAs as they have never liked direct pricing comparisons and will question their added value over customers booking on their own website.

For homeworkers and high street agents, dynamic packaging fell out of favour during COVID-19 because of the heavy administration burden caused by flight disruption. Many agents preferred to take lower margins but less hassle from selling third-party bonded agents. Still, it will be interesting to see what impact a trade-friendly Ryanair could have on reviving this market if, as expected, Ryanair look for further trade deals on the back of its early success with OTAs.

We may never like Ryanair, but a peaceful coexistence offers many upsides and will increase price competition for the benefit of all holidaymakers.

The War is over and long may it remain that way.