I’ve discussed the ideal Travel AI Service should read my mind and feel like a personal extension of myself.
I would like to visualize what a highly automated, and eventually completely autonomous, Travel AI-powered booking might look like.
I believe everything I describe below is possible today to (i) increase conversion rates, (ii) increase engagement, (iii) increase brand loyalty and differentiation for the travel retailer, and (iv) offer a less friction filled travel experience.
One Simple Search Request.
“I want to attend the SKIFT Global Forum”
Can we pause and fully appreciate the ease of “travel natural language search” needing such a simple starting request, one I hope to justify as possible below.
AI Makes Natural Language Possible
Natural language search has been attempted for years, I know, but AI actually makes it possible and universal to implement.
Traveler Search Behavior Needs and Will Change
Traveler behavior need to change and will change, in a variety of ways for a variety of reasons, away from the current to/from booking engine experience.
This change will take years, and it’ll vary by generation starting with a generation (Gen B? Gen C?) who are born “natural language search natives” (versus Pavlovianly trained generations before used to/from search boxes).
Trip Search
My Travel AI bot identifies city pairs via my travel profile, accessing my Outlook calendar, typically tagged with location and priority, to determine my whereabouts and obligations in the days before and after the event.
The bot is aware of the event’s start and end dates, and the address of the event using LLM Search results from Bing, Gemini, OpenAI, or Anthropic.
Travel AI already, based on my personal Travel LLM history, knows I usually don’t stay for the entire conference week based on a combination of: (i) I don’t have the ending plenary on my calendar, and historically, (ii) I return home from NYC on Thursdays nights 90% of the time to enjoy weekends with my family and (iii) I don’t have event sessions on my calendar.
Inbound/Outbound Flights
In terms of Inbound/Outbound Flights, the bot predicts Sunday mid-day or earlier as the optimal day for outbound flights because my Outlook calendar shows no key personal or professional commitments (my sons football tournament, flagged high priority on my calendar, is on Saturday).
Travel AI also knows, historically, based on my personal Travel LLM, 60% of my outbound flights to North America and 90% of my outbound flights to JFK occur on a Sunday, even when I could have departed on a Friday or Saturday.
My Travel AI bot checks all flight options, taking into account the event’s opening plenary at 9am on Monday, accounting for travel times between DOH and JFK across all airlines, transfers to/from airport, hotel check and it’s proximity to event, downtime so I can rest, and time between hotel and event.
Intelligent Ranking of Options
Travel AI then applies intelligence to rank options based on:
- Price & Value & Time
- Analysis of all my past itineraries (my own personal Travel LLM that I’ll touch upon in another post)
- Consideration of all my preferred loyalty programs, their perks and rules (air, hotel, car, credit card, corporate perks). Travel AI tracks what I get free due to my status, my class of service, so I nor my travel agent have to.
- Application of any applicable corporate policy rules (for example it only makes changes if it costs less than USD 1000 or less)
- Filtering for the myriad of learned and asked HPR’s (Habits, Preferences, Rules”) that apply in general and separately for specific city pairs.
- Leveraging the proprietary recommendation algorithm designed by my OTA/TMC or respective travel tech platform (which learns from not only my earlier trips but others in my company, and anonymously from trips across all their clients).
Trip Booking
After this ‘intelligent AI filtering’, it ranks Qatar Airways from DOH to JFK as the top choice, recommending UA and AA as alternates.
Travel AI knows to use OneWorld carriers (when displaying options, and in booking) given my flight history (in general) and to New York based on my personal travel LLM.
Travel AI automatically chose the correct credit card (one of three) that maximizes my status and points (it analyzed each each credit card’s travel rules and realized my 2nd and 3rd credit cards don’t offer points unless booked directly with the credit card travel department).
On the topic, there’s a whole unexplored area (and highly lucrative) of Travel AI trained in a B2B capacity of intelligent suggestions and adjustments in lieu of the current ‘if-then’ type rules.
One or a Few Results
Travel AI will generate a “best result” — Qatar Airways in this example — and a few additional OneWorld options and intentionally filter out, increasingly more intelligently, the set of results I prefer, currently available and their priority.
The traveler can request additional options, using natural language queries, however the Travel LLM intentionally avoids displays dozens of flight options because many will never be a viable choice (and my travel LLM history confirms this).
Travel LLM Filters like FB Curates Your News Feed
This is the where proprietary Travel AI algorithm decision making comes in just like Facebook decides which of your friends posts you see versus ones you don’t; Travel AI will do the same for the search results.
Day Of Travel
On the day of travel, the Travel AI bot automatically orders an Uber 90 minutes prior to departure having calculated a 15-minute drive to the airport and a 15-minute check-in process, as observed from actual earlier Saturday morning departure itinerary experiences, and for today’s trip via 3rd party API’s (Google Maps Traffic, Airline Offered API’s, GDS Data).
For some airports it would even direct me to the shortest security line and points of interest closest to my departure gate.
During My Trip
During my trip, Travel AI would naturally proactively monitor and handle any flight changes or cancellations and cascading repercussions therefrom.
If my return flight is leaving 2 hours later, Travel AI will request a later hotel checkout and readjust my transfer to the airport.
Travel AI will automatically submit a travel insurance claim for the flight delay compensation.
It actively monitors my location, my calendar and my existing travel itinerary for any unexpected changes during my trip.
Travel AI checked real time traffic data for my car ride to the airport before and during my meeting to warn me to leave earlier or later.
I’m Running Late
On the day I’m returning home, knowing I’m a bit behind schedule, Travel AI suggests I checkout of the hotel and then leave for my final meeting, knowing my final meeting is closer to the airport. Travel AI knows I have a driver taking me to the airport so it’s no problem to request an earlier pick up and leave my luggage in the car, so I can leave directly after my meeting and save 20 minutes. It even sends the chauffeur an SMS notifying him of the situation just to make sure my travel experience is as frictionless as possible.
Easy Changes
Travel AI makes it equally effortless to change my itinerary using natural language commends.
‘I want to leave 2 days later’ initiates a cascading set of changes across my entire itinerary:
- Extending my hotel stay and paying for the extra charges with the card on file. Hopefully my hotel app picks up on the change and extends my electronic app based room key.
- Changing My flight includes my current and comparable (OneWorld airlines in this example). Thankfully the change fee was less than the threshold requiring human authorization.
- My airport transfer.
- My Outlook calendar and my shared family calendar.
- It even prompts my assistant with a list of conference calls needing rescheduling since I’ll be in the air (or should I do them via the QR GX In-Flight WiFi 😉
- Etc..
I think you get the idea.
Intelligent Upselling
Travel AI will also be used for intelligent upselling and promotion pushing at various points in the journey in a proactive and reactive manner, in a sponsored and organic manner.
Lounge Access: I want Travel AI to upsell me a lounge in Doha ONLY AFTER checking all free options (status, class of service, none of my credit cards offer lounge as a perk, I don’t have any vouchers in the system).
Conference Room: seeing I have a call scheduled during my time at the airport (or during a layover), it offers an office I can rent by the hour (in this case I can use the QR lounge but good to know).
Make It a Staycation: since Travel AI knows me, presumably (eventually) even better than anybody else, it will search for discount fares for my wife and children to join me turning my weeklong business trip to now include a weekend staycation, only if there’s a special airfare, visas aren’t required in this case, there’s hotel availability at my current and 3 other family friendly hotels, and if the weather is expected to co-operate — all of which are verified by Travel AI.
I am not doing “upselling” justice but suffice to say Travel AI accelerates existing and entirely new upsell opportunities and touchpoints.
Doesn’t this increase revenue & traveler satisfaction?
- Will this yield higher booking conversion rates? I believe so.
- Would the traveler experience would be better or easier? Definitely. No trip is drama free, but I believe Travel AI would make the traveler’s (and their travel agent’s) experience considerably easier.
- Does all this require historical, current and predictive data? Absolutely! My personal Travel LLM is trained, tweaked and maintained (by your travel tech provider, GDS).
- Is what I’ve described impossible? No, and well worth the ROI required.
- Is it easy to implement? No, however it is possible given current 3rd party API’s, 3rd party integrations and the correct AI training data.
This “Travel AI Trip” Is Already Possible
I believe the above is possible today, actually not particularly difficult in many aspects, with new and existing travel technologies.
It’s not a weekend or 72 hour hackathon project but it is within possibility for a travel startup with no or little funding.
My Example Scratches the Surface
My example is just scratching the surface on what’s possible — they are by no means anything even close to the ideal Travel AI booking possible today (that I’m aware of and if you know, do share!).
Share Your Thoughts.
PS, I’d like to think I’m not just some rose colored glasses believing the AI hype. My conclusions are based on looking at the current AI API’s available including evaluating their capabilities with public and private training data, knowing the 3rd party and GDS data available, and leveraging my experience as a former boutique OTA owner, travel tech developer and now travel investor.
