Upgrading AI-powered travel products to first class
In the race for dominance in the AI travel industry, even a small lead matters right now.
Every player in the space trying to capitalize on the promise of new AI/LLM (large language model) technologies is struggling with the fact that major platforms like ChatGPT are limited by data that is outdated or not real time. In an industry like travel, where fickle plans and itineraries literally change with the weather, this is particularly problematic.
As both investors in AI travel and advisors coordinating deals for startups in this space with other investors, we like to see companies pushing boundaries and providing value for users in new, concrete ways.
For instance, Kayak and Expedia have launched ChatGPT plugins, but GuideGeek from travel publisher Matador Network provides real-time flight data (GP Bullhound has provided financial advisory services to Matador Network). Meanwhile, Roam Around has a strong visual element to its interface.
But travel information is complex, and incorrect information — or AI “hallucinations” — are a challenge. Roam Around sometimes recommends one site while showing a photo of another (and potentially linking to a third), and in one of our queries, GuideGeek conjured a cleverly named pub that simply doesn’t exist.
Other than Airbnb, there really hasn’t been a major shift in how we plan and book travel online in decades.
At this early stage, our firm and other investors in the space we work with don’t have an expectation of perfection. The advances between GPT-3 and GPT-4 are so apparent that it’s easy enough to look at the underlying technology and say, “Eh, they’ll figure it out.”
We’re more focused on how companies are shaping and augmenting this technology for travelers and the market segments within the travel industry they are positioned to capture.
Differentiation is key — you need more than a skin for ChatGPT
Most AI products seem to be built on ChatGPT. While each travel company may start with the same baseline, we really like to see proprietary data that can train the AI to produce superior outputs. OTAs (online travel agencies) like Expedia and Booking.com have an advantage here with massive amounts of information about their customers and how people plan and book trips.
Small agile teams have an opportunity to adopt this technology and rapidly scale up before the big guys can effectively implement or risk disrupting their existing business. If consumers can use an AI tool to search all airlines instantly, why does an OTA need to be in the loop?
The OTAs are built on recommending what the masses want, but the whole point of AI is that the answer is now customized to the individual. An average ranking of 8 for a hotel doesn’t apply to a specific person whose main priority is to be close to a lesser-known local surf spot an AI surfaced for them.
To try to drive utilization, startup players have to get more creative with product design. Getaiway and Roam Around have focused specifically on itineraries, with the latter simplifying the user input down to one word — type in a place and get an itinerary, then refine from there. Matador says it plans to include influencer videos from its wide range of content creators in the GuideGeek messaging interface.
Upgrading AI-powered travel products to first class by Walter Thompson originally published on TechCrunch
