Got a problem AI might be able to solve? Let me introduce you to the most likely solution.

Over the past 12 months I have probably made more than 150 introductions for people who reached out with an AI problem. Someone needed help, I knew someone who could solve it, and an email got the two of them talking. Almost all of them are readers of this newsletter.

As a startup founder, I also know that finding the right customers is still one of the hardest things to get moving. There are good people building useful things. There are travel businesses with problems those things could solve. Getting them into the same conversation should not be this difficult. But it is.

So far, most of those introductions have been pretty serendipitous. An email lands. Someone sends a LinkedIn message. So I’ve been thinking: Why should the benefit only go to the people who had the confidence to reach out?

So we have built something new for the Everything AI in Travel community: a way to connect a real travel business problem with someone who can actually help.

No need to know the right AI terminology. Simply tell us in plain language who you are, what is frustrating you, taking too much time or costing too much money, and how you are handling it today. Plain words are perfect.

If there is a strong fit in the network, I will offer to introduce you personally to the person running that business. It might be someone you have heard on the podcast or someone in my wider network. What matters most is that they can most likely help.

Not a random list of websites. An introduction to have a conversation with the most likely right person.

It is free for those who subscribe to the newsletter and listen to the podcast (that’s you BTW).

Tell me what you want AI to help you solve. I’ll recommend the person with the solution.

And if you are building an AI solution for travel, I want to hear from you too. What do you solve, who is it for, who is not for and what is working in the real world? I’ll try to connect you with your ICP.

If you are building an AI solution for travel, tell us about it here.

Southwest moves the front door. Keeps the till.

A week ago these ChatGPT apps built by customers were laughable. A customer had to first install it and then actually call it in the prompt they were giving for it to be used. It was an ugly duck. 🦆 Now those ducks just might be beautiful swans 🦢

Southwest has moved its shop window into ChatGPT. The till is still on Southwest.com That distinction matters more than the “AI flight booking” headline.

On October 6, Southwest launched a ChatGPT plugin with OpenAI and AWS. Travellers can discover and shop for its flights in the conversation, then move to the airline’s website to complete the purchase. Robert Silk at Travel Weekly reports that the flight information comes directly from Southwest’s systems, rather than being reconstructed from public search results.

The commercial play is pretty clear. Southwest wants to turn up earlier, while the traveller is still working out where to go, then bring the purchase back to its own site. Direct booking does not have to start on your own website anymore.

That does not mean the OTAs are finished but this also doesn’t help them.

In last week’s Everything AI in Travel newsletter we talked about how OpenAI showed that the system itself will pick relevant Apps and run them if it improves the answer to the question. In travel that is almost always the case. Being able to harness from first party data allows real pricing and availability to shine through.

We do not have conversion numbers here, and this is a first phase, not an autonomous agent buying tickets while you sleep. But it does mean suppliers have another possible route into the customer’s consideration set.

Radisson brings a conversion number to ChatGPT

Yesterday Southwest gave us a new shop window in ChatGPT. Today Radisson has brought a number to the conversation.

Radisson says its ChatGPT plugin is converting at around 1.5 times the rate of its organic search. That is rather more interesting than another announcement that a hotel company has embraced AI.

OpenAI published the case study on October 7, with Radisson and Accenture Song. The app lets travellers compare hotels, prices and amenities inside ChatGPT, then sends them to Radisson's own website to finish the booking. The app itself launched in July. The news here is the early result, not a brand-new launch.

A useful caveat before everyone starts reallocating their entire marketing budget. This is a company-reported result in an OpenAI customer story, not an independent test. We do not have enough detail on volumes or comparable audiences to say the plugin caused a 50% uplift in bookings. A higher conversion rate is not the same thing as more bookings overall.

Still, for mine, this is the bit that travel companies should be looking at. Radisson is getting a customer from a conversation into its direct booking channel. It does not have to own the place where that conversation starts.

There is another change worth noting. OpenAI says people who have the Radisson plugin installed no longer need to type @RadissonHotels to call it up. ChatGPT can show it when relevant. Installed is the important word there. This is not every ChatGPT user suddenly being shown Radisson.

Radisson is also running sponsored advertising in ChatGPT. So the new discovery channel already has both an organic path and a paid one. That should sound familiar to anyone who has spent the last decade paying to be found on Google.

The OTAs are not finished. But a hotel brand connecting its own rates to the conversation, then taking the booking itself, gives them something else to think about.

If you run a hotel, I would be asking your team how you measure AI traffic beyond "we got some clicks". Does it book? Does it bring a new guest? What did you pay to get them?

Google puts hotels into UCP: channel or toll booth?

Hotels just got a seat at Google's AI booking table. Is it a new channel, or a new toll booth?

Lodging has been added to Google's Universal Commerce Protocol (UCP). A September 25 announcement opened the draft spec for public feedback, co-developed with a Lodging Tech Council that includes Booking.com, Expedia, Amadeus, Hilton, Marriott and Trip.com. Google's pitch: direct bookings inside its AI surfaces, including AI Mode in Search, with the hotel staying merchant of record.

That "direct" is doing a lot of work. The guest may be booked without leaving Google, but the hotel still isn't the one who owns the moment of discovery. It's a distribution channel, and the question for hotels is what it costs to acquire that guest and whether they ever come back to book direct next time.

The devil as always is in the detail. The draft says the full stay price has to be shown before confirmation, including mandatory charges collected at the property. Contributors are already asking what happens when a request times out and the guest finishes on the hotel's own site (two bookings?), and what happens when cancellation terms change but the price doesn't. Same room, same rate, different purchase.

It is still a draft with no consumer rollout date. But the Booking.com and Expedia names on it tell you the OTAs are not sitting this one out. They plan to be inside the protocol rather than outside it.

Personal agents need a front door into your business

Meta and Sierra want to give personal agents a front door into every company. For travel, that could change how customers book, change and manage a trip.

Personal Agent Protocol (PAP) is an open standard being developed by Meta and Sierra with Genesys, Rocket, Shopify, Stripe, Walmart and Instinct. It sets how a personal agent talks to a business: start as a guest, sign in when the task needs an account, and work through the company's website, its APIs (MCP, OpenAPI) or the company's own agent.

Today most agents just click through websites like a person, or end up on a support line. That is slow and it fails. A direct connection does the same job in seconds.

Two things stand out. The customer decides whether the agent gets read-only or write access. And the company decides what the agent is allowed to do, and gets to see when one is acting for a customer. That visibility is the commercial point. Brands don't want agents going round them, they want to know who's knocking.

The next extensions listed are the interesting ones for travel: push notifications so a company can tell an agent the moment a flight is delayed, and payments that let an agent buy without ever seeing a card number.

If you think someone (or everyone) you know or work with could grow from being more informed on the topic of ai + travel (or could use the training above) then please forward this email to them and they can click the button below:

AI gets the credit. Your booking site gets the blame.

Propellic watched US travellers plan real trips with their own AI tools for a week. Screens recorded, thinking out loud. More useful than another survey asking whether people reckon they'll use AI.

71% were pointed towards booking direct with the operator. Nice news for hotels and tour operators. Less comfortable for the OTAs.

But more than three quarters of booking handoffs failed to carry across dates and guest counts. Travellers blamed the booking site, not the AI. Ouch. The AI can drop the details and you still wear it.

Two other findings caught my eye. Most travellers stuck with one AI tool rather than checking another. And when asked why they trusted recommendations, they pointed to comparison tables and ranked lists. Presentation was doing some of the persuading.

For mine, this is a direct-booking opportunity with a fairly unglamorous catch. Getting into the answer is only half the job. The next step needs to work, even when someone arrives without the context they already gave the AI.

Before buying more AI visibility, try planning a trip to your own business. Follow the recommendation through to booking. How much do you have to tell it twice?

Note: This is an observed study, not a forecast for the whole market. But it's a useful test to run now.

Slack Group!

Podcasts and Sponsors

Podcasts now on Spotify and Apple Podcasts:

All podcasts are now showing up on Spotify and Apple Podcasts for your easy listening pleasure! Now is a great time to binge those you may have missed.

The latest released episode is my conversation with Emmanuel Annor, co-founder and CTO of Equera.ai.

His starting point is simple: AI cannot recommend or book experiences it cannot see. We talk about bringing community-based tourism into the digital travel economy without asking small operators to abandon WhatsApp, paper itineraries and the way they already work.

Listen to the episode here.

Partner with Us - sponsorship info for the newsletter and podcast can be found here.

Most clicked last week was Rafat Ali's video with Noah Shinn about Instinct.

That is it - you have made it to the end of this edition. If you are thankful for this newsletter, you can buy me a coffee.

If I have missed something, you have a tip or any feedback at all, simply reply to this email and it will come straight to me. I am doing this for you, so please do not be shy about telling me what you think.