Leakage has historically been one of the critical problems affecting business travel booking.

Whether it is because of the superior UX of consumer sites or because of the restrictions imposed by travel policies, business travellers have increasingly skipped official booking channels.

Besides other things, leakage has made reporting less accurate (how much are we actually spending on hotels?) and duty of care more difficult (where are my travellers?).

Generative AI could completely revert this trend. 

Because of their need for structured data, booking tools have historically required travellers to be very precise when submitting a request. 

Things like origin, destination, round trip (yes/no), day, time range, preferred seat, preferred airline have been the core ingredients of the ultra-sophisticated (and expensive) booking tools the industry has built in the last 25 years.

When GPT-4 was launched a few months ago, this constraint started to disappear.

Thanks to the latest generation of LLMs (Large Language Models), we have unlocked one superpower that felt science fiction just a few months ago. We can now make computers understand and extract meaning from natural language. 

The times of “Sorry, I couldn’t understand your request” that we suffered in the previous generation of chatbots is over. 

Models like GPT-4 can help us extract meaning from:

• Direct traveller requests → “I want to change my flight later today”
• Traveller interactions → Shared calendar appointments, apparently unrelated conversations (“My last meeting has been cancelled”), traveller historical booking data…
• Any other traveller/company related data → company travel policy docs, traveller preferences registered in the traveller profile.

What does it mean for the future of business travel booking tools? 

I think we will see a gradual change from “travel initiated requests” to “AI suggested options”. 

“We have unlocked one superpower that felt science fiction just a few months ago”

Thanks to all the context that LLMs will be able to collect and analyse about the traveller, AI models will be able to generate increasingly more relevant suggestions. 

Personalisation won’t be just a “keyword” anymore. 

What the AI will be able to suggest to the traveller will cater, at the same time, to what the traveller wants (preferred airlines, type of seat, departure time based on historical data…) to what the company needs (selected vendors, cost restrictions, CO2 impact…).  

Relevance (what the traveller wants and what the company needs) will be combined by ease of use (how the traveller books). 

It takes 10x less time accepting a relevant trip suggestion by typing “Yes” to a Slack message that an AI just sent you, than to initiate a booking request from scratch by using an OBT.

When we combine relevant suggestions with frictionless UX, we get traveller adoption. 

The change will take time. The examples discussed above still require a lot of work, but the ingredients to make them a reality are already here. 

Generative AI could help the business travel industry solve travel manager needs and fulfil traveller desires in one single shot.

Cesc Vilanova has over 20 years of experience in building digital products. Before entering the travel industry, he spent a decade working as a UX designer and creative director for various gaming startups. In 2014, Cesc co-founded Sam :] , an award-winning digital assistant acquired by FlightCentre in 2017. Following the acquisition, he assumed the role of Chief Product Officer (CPO) for FlightCentre’s business travel division. Since mid-2021, Cesc has been working as an independent consultant, aiding travel and e-commerce companies leverage new digital technology to create new digital experiences.