How AI-based personal assistants are very likely to change the way we book a hotel, how distribution will most likely be affected and what action hotels should implement now, to avoid missing out on the revolution, which has already begun.
If we are not mistaken, we can already imagine that we will all have our own personal assistant, a digital twin, let’s call it Sammy, (provided by Amazon, Apple, Google, Meta, Microsoft, Open AI) in the palm of our hand, on our smartphone, or even sitting right on the tip of our nose, with connected glasses, acting as digital concierge.
What may have seemed like science fiction a few months ago, is increasingly becoming a tangible future. Let us now project ourselves into the field that interests us, the hotel industry. When, why and how will we use Sammy?
Chronologically:
To find a hotel according to simple criteria such as location and price, OTA platforms (such as Booking or Expedia) remain essential. It is therefore very likely that Sammy will start its search on one of these platforms. However, as is often the case today, after selecting a few hotels through these sites, we would usually consult the hotel’s official website to get more information or better understand its ‘ambience’. In the same way, we can imagine Sammy asking the hotel’s AI virtual assistant (like Velma by Quicktext) for additional information he needs. In this field, Anthropic has established an open-source standard, the Model Context Protocol (MCP). This protocol enables connecting databases (like Q-Data) to AI models and personal assistants. This way, they will have direct access to relevant data, allowing for intelligent and contextual responses, thereby enhancing the efficiency of AI.
The information that Sammy will have access to, will be what is publicly available on the web, such as the hotel’s website or with the OTAs. In other words, this data will be limited and will represent, at best, only 30% of the full knowledge, as explained in the article: Sophistry & Artificial Intelligence. Therefore, the risk of errors or hallucinations will probably be very high, even with the new generative search models. In short, Sammy will only be able to answer very simple questions, and it will be essential to check the answers. And for this, as mentioned above, Sammy will need to turn to Velma to obtain reliable and detailed information.
Sammy will probably be able to count on the OTAs for his research. But then, which one? You can guess what happens next: each OTA will do everything they can to become Sammy’s default booking system. And what about hoteliers? If they have a virtual assistant like Velma, they will be able to compete (by competing with OTAs). But what about those who don’t have a virtual assistant? The answer is simple: sadly they will be excluded from the game. They will be on the same level as a hotel which doesn’t have a website today.
In addition to offering rooms, virtual assistants will also offer local experiences, such as tours and events, in collaboration with local partners. This will make the offering much more varied and attract a clientele more eager for personalization, thus also reducing the interest in the OTAs.
The only solution would be for Sammy to be connected to all hotel reservation systems, which seems unrealistic. On the other hand, we can imagine that he acts as an intermediary by putting us in contact with the hotel. However that will mean that hotels will be overwhelmed by millions of requests sent by Sammys from all over the world. Do you see the problem? Hotels will absolutely have to have an AI like Velma to manage these requests coming from assistants like Sammy. And, as Simone Puorto says, in a world where the number of tourists is constantly growing and the number of people wanting to work in the hotel industry is constantly decreasing, human intervention will become a luxury.
( Read Simone Puorto’s article: The impact of AI in the Travel and Hospitality sectors ).
At this stage we can therefore see the interests of the parties involved.
The interest of the user
That of the hotel
That of the personal assistant
The hotel’s virtual assistant has non-public information, called private information, which was integrated during its configuration. On the other hand, the personal assistant only has access to public information and cannot access this private data. This private information therefore has real value, because it helps attract traffic and generate booking requests.
We come to the first conclusion: the hotel’s data will be worth its weight in gold. But on five conditions:
This will be the role of the CKO (Chief Knowledge Officer), to ensure that the data that feeds the AI meets these 5 criteria.
In addition, it is easy to imagine that this data is much more reliable than the data collected by the OTAs over many years (full of different information, opinions, messages, reservations).
Indeed, this data does not meet the criteria:
Can someone explain to me how such a combination of approximations can generate relevant answers?
(Discover the challenges of Generative AI in hospitality, why ChatGPT-based projects fail, and the risks of AI-generated errors in ‘Sophistry & Artificial Intelligence)
Sammy will start by connecting to the web to find basic information about the hotel. If we want him to respond quickly and accurately, we will have to ensure that this information is widely disseminated and correct. To do this, hotels, just as they use a channel manager to manage their prices, will have to have a “data channel manager” (like Q-Channel). The latter will have the mission of providing the OTAs, social networks, and other platforms with certified data about the hotel in real time.
What would be the danger in hotels distributing their data, especially to OTAs?
What would prevent OTAs or even virtual assistants from recomposing the data?
I will quickly go over this point, because it is blatantly obvious. If your personal assistant is going to talk to the hotel’s virtual assistant like Velma, the latter will need to be connected:
The latency of a conversation between two humans is 3/10 of a second. In the context of an AI, we are looking at more like 3 seconds. Suffice to say that the conversation will not be very fluid. For this, we can always count on technological progress, but it would be more appropriate to work with a structured database such as Q-Data rather than with a huge data lake grouping together disparate data, the processing time of which will de facto be longer. Indeed, the energy consumption of AI is already a major problem and making them frugal is a future challenge.
Indeed, communications will be omnichannel (Read our article ‘AI in Hospitality‘ by Benjamin Devisme to explore the challenges and realities of AI in the industry): they could start with voice, continue via a screen, then end with a chat exchange. Imagine for a moment having to dictate your last name or email address to an artificial intelligence… Good luck! It is therefore very likely that innovative solutions will emerge, such as sound signatures. Imagine a sound equivalent of a QR code, integrating data such as name, email, telephone number, address, an identifier, or even bank card information.
Even better, Sammy could already memorize this information and transmit it directly to the hotel’s virtual assistant. I will briefly go over the issues of confidentiality and security, which will undoubtedly be addressed through solutions based in particular on blockchain and ZKP (Zero Knowledge Proof).
The probable arrival of personal assistants (such as Operator by OpenAI) will change the way we communicate with hotels. Indeed, these assistants will gradually take over. So for hoteliers, it is a question of preparing now by adopting a virtual assistance solution (such as Velma) that will be able to communicate with them.
The four conditions (KIDS) to be met now are
If hoteliers are wise enough to follow this plan, which is based on simple common sense, they will be able to lay the foundations necessary to take full advantage of the revolution underway. Personal assistants will redefine the hotel distribution landscape by limiting dependence on search engines and OTAs. By offering more control to users and hotel chains, these assistants could create a new direct and personalized distribution channel. Google and OTAs might even see their current role slightly diminished. To remain relevant in the face of these new assistants that will simplify bookings, they will need to adapt and rethink their services. However, don’t worry about them: with their history, expertise, APIs, and substantial marketing budgets, they will know how to capture the flows of AI agents and maintain a dominant position for many years to come.
The future is (already) now!
© Image: Shutterstock
