Key Takeaways:
Google is enhancing its trip planning capabilities to offer a more comprehensive, seamless, and personalized travel experience.
The new service is significantly more advanced than Google's existing offering but still lacks several basic features available through today’s offline agencies.
This heightened level of service is achieved by integrating Generative AI (GAI) with a new technology called AI agents.
Future announcements are likely to address remaining gaps, creating a more comprehensive and superior agent service.
This development could significantly impact today’s travel sellers (especially OTAs) and Google may not be the only large AI company developing travel applications.
Note: To avoid confusion between travel agents and AI agents, whenever we use the term ‘agent’ alone, it refers to an AI agent.
On May 14th, Google unveiled a new version of its GAI-powered travel capability, called the Trip Planning Experience in Gemini Advanced. This new trip planning experience builds on Google's existing flight, hotel, and general search capabilities with several new features:
A text-based conversational interface that allows travelers to describe their intended travel plans as they would to a human agent. No more filling out extensive widgets with specific instructions that are likely to be ignored in the search response.
Recommendations for entire itineraries delivered in seconds based on natural language instructions and preferences, considering factors like seasonality, travel time, and probability of fatigue after each activity.
Personalized recommendations for places to visit and things to do, based on user requests and past preferences, prioritizing customer preferences over seller margins. (Imagine that—traveler personalization above profit optimization!)
Flexibility to accommodate changes in plans, adjusting the itinerary as needed, and in just a fraction of the time needed via other means.
Despite these advancements, some common capabilities of traditional agencies remain unmatched (or unmentioned) in the demo:
The demo assumed previously booked flights and hotels, with no example of the booking process, though future booking capabilities were mentioned. In the demo. The agent retrieved flight and hotel booking data from the traveler’s email.
The conversational component is text-only, whereas offline agencies offer Zoom calls or in-person visits, which provide a more intimate relationship between agent and traveler.
Price shopping and monitoring.
24x7x365 customer support.
Ancillary services such as insurance, price protection, and flight interruption handling.
AI Agent Technology: Google's new capability is an example of an AI Agent interface, an emerging AI advancement we’ll all be hearing a lot more about. Unlike today's chatbots, which use a single generative model to accomplish a specific task, agents can use multiple models and data sources to achieve goals using planning and reasoning capabilities. While still imperfect and fallible, these agents will improve significantly over the coming year. Combined with multimodal models that use text, vision, images, sound, and video, their capabilities will include real-time, lifelike human conversations in the not-too-distant future.
The planning capability in Google's agent involves reasoning about logistics, prioritization, and decision-making. It gathers information from (Internet) search, maps, and Gmail, creating a dynamic graph of travel options that considers both the traveler’s priorities and constraints. If inputs change, the agent recalculates and searches the graph again for the best recommendations, a process that might take hours or days for a human agent but is completed in seconds by the AI.
Additionally, the agent offers real personalization, and this deserves a little deeper dive. This personalization is based on:
User requests and subsequent instructions.
Past inquiries and behaviors (Google now retains search query information for up to 3 years by default). Travelers only need to tell the agent your preferences once, or mention you’re an avid scuba diver, are a vegetarian, have a fear of heights, prefer public transportation, or whatever, and have it remember that when building recommendations every time—unless you change those instructions.
Current trends and environmental conditions. Looking to go to that new destination? Did you know about the wildfires there? Or that volcano?
Collaborative filtering insights. In other words, you really liked your trips to x and y, and other people like you who also liked x and y liked z as well, so why not consider a trip to z?
Customized pricing and offers from suppliers. Many suppliers are keen to personalize offers to customers and potential customers but just don’t do a very good job. (Why do you ask me at check-in if I’ve stayed here before? I’ve been here a dozen times!) If the agent can perform the personalization based on a variety of inputs, the suppilers only need to say what they would like to offer to customers who meet a particular spec.
Is your future agent human, or just human-ish? While the demo didn't show equivalents to all the great features available through traditional travel agents, these gaps will likely be addressed quickly. Even the human touch in conversational travel agents can be replicated with new technology. For instance, Microsoft's Vasa-1 technology can create a photorealistic avatar from a single photo and a sample of a human voice, enabling real-time conversations with a lifelike agent. Those future zoom calls with agents who are available 24 hours per day and never lose patience are on the way.
New Agency Business Models: Because technology can replicate travel agent capabilities at a very low marginal cost, Google could introduce new agency business models. GAI travel agents could change the business dynamics in several ways:
By using GAI travel agents and customer service capabilities, Google could significantly reduce their cost to serve, enabling a subscription model rather than relying on transaction margins. (Which enables that hyper-personalization we discussed earlier...)
Offering continuous travel inspiration options for various customer needs. Put another way, they could maintain the current model where agents wait to hear from travelers about their desire to travel and include other options where travelers can make general plans for a year’s worth (or more) of travel in advance to take advantage of lower prices, limited availability of popular activities, well defined vacation time periods, etc. It might include options for:
Families with school-aged children could input school holidays and receive integrated travel plans for the year, including spring break, summer vacations, and holiday visits.
Price-sensitive travelers might provide a list of potential destinations and times, letting the agent construct and optimize travel plans based on current deals and monitor them for price dips.
Intergenerational travel, which is growing in popularity, could be managed more efficiently with larger planning graphs tailored to diverse age groups. Larger groups can be an absolute nightmare to manage—for humans—because of all the constraints and options that have to be managed across a larger number of travelers. It would be far easier to manage these extensive travel programs with GAI.
By building travel-savvy foundation models and agents, companies like Google can use the most advanced GAI technology and amortize the high fixed costs over a large user base, resulting in low marginal costs to serve. This allows rapid development and deployment of more advanced capabilities--with global reach.
This also points to the reason why OTAs might be more vulnerable to companies like Google than offline agencies. The advantages OTAs have used since the late 90’s to leapfrog over offline agencies have to do with using large amounts of data and the smart application of technology— the very things companies like Google, Microsoft, and potentially Apple can use with their advanced GAI to leapfrog over the OTAs. Offline agencies, on the other hand, leverage the human element in the relationship with travelers, with agents often sharing their own experiences in traveling to places they recommend. Building rapport with clients and sharing small details about the contents of a room at a hotel or the best coffee they discovered on a trip is still out of reach, and likely will continue to be so, for even the best GAI models. Thus offline agencies that leverage GAI for its abilities to reinforce what human agents can offer is likely to be best overall solution for travelers.
Emerging Competitors: Google may not be the only player in this space. Apple and OpenAI have reportedly completed a deal to use OpenAI technology on iPhones. While neither has demonstrated anything like Google’s new Trip Planning Experience, they have a lot of the ingredients to do so. OpenAI continues to have one of the most advanced models with GPT-4o, and the upcoming GPT-5 is rumored to be far better in all dimensions, including the ability to plan and reason, and it will also have the ability to gather personal preferences and create a traveler profile. Apple brings billions of devices, a premium customer base, and mapping data that includes incredibly rich descriptive content for destinations, hotels, activities, etc. The combination of these two companies could provide some additional competition in the travel distribution industry.
Conclusion: The integration of AI and agent technology is revolutionizing travel planning, offering personalization and efficiency like we’ve never seen before. While Google leads the charge with its new Trip Planning Experience, the landscape is rapidly evolving, with potential competition from tech giants like Apple and OpenAI. The future of travel planning is poised to be more dynamic, personalized, and competitive than ever before.
Author: George Roukas
George is a senior executive with in-depth experience in product management, technology, and competitive strategy. George was a co-founder of Hudson Crossing, LLC, a management consulting company dedicated to the travel industry, in 2007. Prior to that, he was Group Vice President of Product Management for Travelport, where he led the strategy, development and management for all products facing Galileo’s North American corporate and leisure agency partners. He also participated in the leadership of several early e-commerce companies including Biztravel.com, Room12.com, and Clickradio.com.
At the end of 2022, convinced that new generative AI models coming out were going to be unprecedented in their impact on nearly everything we do, George left Hudson Crossing to study AI, specifically in its applicability to business, full time. He now advises companies how to best adopt generative AI through his new company, GAIPAN.
Comments