From AI-driven chatbots to the arrival of blockchain-based cryptocurrency fuelled competitors to Airbnb, last year was a pivotal one, says Mezi
In 2016, Mezi, a startup that originally launched as a broad-based AI-power virtual assistant pivoted to focus on travel. EyeforTravel caught up with Johnny Thorsen, the company’s VP of Travel Strategy & Partnerships, who believes that 2017 was the most important year since the arrival of the Netscape Internet browser in 1994.
EFT: That’s quite a big statement. Why do you think 2017 was such an important year?
JT: Until now the travel industry hasn’t been able to take advantage of the unlimited number of customer interactions processed via online travel solutions. In fact, all we knew about a user when they booked via a browser-based or mobile app based form is who they were and what they bought (or at least searched for).
Imagine if we knew their emotional state. Imagine we were able to detect the level of intent to buy vs. search, based on the words used in an actual conversation. This is the new world of opportunities we have opened up in 2017 through applications of AI-chatbot technologies.
Despite the arrival of smartphones ten years ago we haven’t really changed the process and workflow associated with travel search and booking. But the combination of extremely cheap cloud computing power and AI-chatbot technology makes it possible to finally create a truly personal end-to-end travel experience.
At the same time, the emerging usage of blockchain technology will enable new consumer friendly distribution services capable of offering search and selection options not available in, for example, the traditional global distribution systems (GDS) or travel management company (TMC) based channels.
EFT: Before we talk about blockchain, which certainly dominated headlines last year, let’s talk a little more about AI-fuelled chatbots. What is their role in the traveller journey?
JT: Travel planning lends itself perfectly to the automation and personalisation that artificial intelligence and machine learning can provide. AI can help understand the traveller’s preferences like a human travel agent can, while machine learning can help fine tune the most personalised recommendations based on the traveller’s preferences.
Additionally, AI can help make human travel agents more efficient as they work hand in hand with chatbots - allowing companies to increase their productivity and utilisation by as much as five times.
EFT: But not all chatbots succeed. At our recent Amsterdam event, for example, Iceland Air admitted their chatbot had been something of a white elephant…
JT: Since AI is a fairly new concept, there will be scenarios in which it fails to understand the travellers need accurately and assumes the wrong preferences. When this happens, it's important to seamlessly transfer the request to a human travel agent, so that the service never fails, even if the AI fails. Currently with Mezi, 60% of requests are handled without human intervention and we hope to shift that to 80% looking ahead to 2018.
Today 60% of Mezi requests are handled without human intervention and we hope to shift that to 80% looking ahead to 2018.
Ed’s note: Just for the record, business travel app Lola Tech, the brainchild of former Kayak co-founder Paul English, which is also coming to the rescue of road warriors maintains the importance of human intervention.
EFT: So, all chatbots aren’t equal?
JT: Chatbots come in two forms. There is the basic version where a number of responses are predefined and triggered by the use of specific keywords or terms. Then there is the advanced version using AI where the responses are dynamically generated based on an advanced analysis of the semantics and sentiment of the conversation.
Think about the concept of an interactive voice response system (IVR) - very few people like the idea of having to say a certain word in a certain way to get to the next menu option. But we all like talking to and being understood by a person - the basic chatbot is effectively your IVR system while the AI-powered chatbot is the human equivalent.
The basic chatbot is effectively your IVR system while the AI-powered chatbot is the human equivalent
Once users realise that they actually get a more personal service based on past behaviour and decision-making, they will find it increasingly hard to go back and use a service that doesn’t offer this capability.
EFT: Where, or for whom, will chatbot technology make the biggest impact?
JT: Business travel is one major opportunity. By some accounts, chatbots will represent 20% or more share of the corporate booking world by 2020. Certainly, frequent business travellers and millennials are drawn to booking tools that utilise AI and machine learning because they provide a natural end-user experience - the convenience and immediacy of the process are facilitated through a text message-style interface, which is how most people are used to communicating on a day-to-day basis.
EFT: So back to blockchain, the big news of 2017. In your view, how far off is it from making an impact in the travel space and where is that likely to be?
JT: Yes, blockchain is making strides, but it is still one to two years away from impacting the travel industry in a meaningful way. Still, I believe that blockchain-powered smart contracts combined with AI-chatbot technologies will allow us to imagine and develop an entire new way of distributing and consuming travel.
With respect to early applications, the technology for travel solutions will likely evolve into a number of private blockchains where the content is made available for others to develop access to. Given the high level of ‘people security’ related to consumption of travel (ID, money flow, airport access, etc.) it makes sense to create a private blockchain which then is trusted and accessed by all relevant parties involved in delivery of the travel experience.
We are already seeing the first [crypto-based] Airbnb competitors arriving on the scene
This is obviously a great opportunity for the big ‘cloud service providers’ which are ready to deliver low-cost infrastructure that could replace the current expensive GDS infrastructure at some point in the future.
Regardless of who the future infrastructure providers are, the cost of distributing travel services will continue to fall with the constant emergence of new solutions.
In fact, we are already seeing the first Airbnb competitors arriving on the scene such as CryptoCribs, which only accept cryptocurrency payments while lowering the cost of both host and guest.
October 2018, Las Vegas