MCP's now make Claude really personal in travel
Hi I’m Tony. My background is from Intrepid Travel where I was inducted into the company Hall of Fame for outsized contribution to the company. Since late 2022, I’ve been going deep on AI + Travel. I share some of what I find here each week and interview people who are building cool AI things for the industry on a podcast.
I also partner with CEO’s, Founders and their boards on making sense of the opportunities with AI in their companies.
AMEX GBT and Reely now both work directly from your Claude account
News this week that those who have their business travel managed through AMEX GBT can now just work with their usual Claude account to make changes or book new things in.
On the consumer side, Reely lets you neatly compile all your travel inspiration on a trip-by-trip basis and then allow you to just chat to Claude as you go to refine, finalise and book. In this case, Claude uses what you have saved in Reely as the basis for what it recommends - not the same generic guff it recommends to everyone else.
I wonder if this is how things might go from here? You have your favourite all-rounder AI (Claude, ChatGPT) and you just work with it on everything?
Both these solutions are using purpose built MCP servers that are now super simple to connect to your personal Claude account (you don’t even need a paid account). It is actually one of the menu items in Claude’s settings (and if you can’t figure it out, Claude will actually do most of the work itself in the set up).
On the Reely side, what this might mean in practice is that users don’t really ever open the Reely app on their phone after the initial short set up period. Instead, they simply save the things per trip that inspire them from their social media feed to Reely, Reely converts these videos into lists of desired products + locations which become the “brain” that Claude pulls from to deliver information back to you.
AMEX GBT have done this work “with an Egencia AI connector integrated into Claude, Anthropic's AI assistant”.
Whilst the article about AMEX suggests “This connector relies on a proprietary architecture known as "agent-to-agent" (A2A): a reasoning layer that breaks down a travel request into tasks (flight search, hotel search, booking) and distributes them to specialised agents,” Evan Konwiser, Chief Product & Strategy Officer at Amex GBT, warns this connector is just one step: " This is a foundational capability, and it's only the beginning"
From the Reely side what this means is that whilst Reely is ready to take your list and negotiate pricing in true A2A style, there isn’t a supplier side tech yet ready to have those negotiations with. I wonder who will fill this space first, a B2B OTA like Expedia of HBX, a legacy GDS like Amadeus or are scale up like Nuitee? It will be interesting.
Of course, when this does come online, the user doesn’t need the detail - they can just keep chatting with Claude to get the job done. Eventually all these companies will likely have this capability and the user can tell Claude who it does or doesn’t want to buy from (if they care).
This differs significantly from “apps” in ChatGPT mechanism that no-one I know is using despite companies still spending to be in there (more below on this).
If you use Claude regularly and want to see this flow for yourself, you can download Reely from Apple App Store, hook up the MCP to your personal Claude and then just save what inspires you to Reely. It’s all free to use and try.
This content is provided by the newsletter sponsor Reely.Travel
AI - its starting to get personal
Travel is throwing billions at AI. But a lot of the industry still seems to think the magic is the interface.
Gee Mann at Business Travel News explains that if every platform is building an agent, but each one is still trapped inside broken, incomplete, permissionless data, then we are just automating the same fragmentation problem at a bigger scale.
That bit feels right.
The examples in the piece are simple but important. The agent books the wrong hotel because it does not have the right event context. It picks a fare outside policy because the entitlement lives somewhere else. It misses a visa issue because the passport data is not there. In other words, the weak point is not necessarily the model. It is the plumbing.
Travel has been here before. Distribution needed the GDS. Payments needed shared rails. AI probably needs its own neutral switching layer too, or the biggest platforms will just build their own memory gardens and call that innovation.
It's not just who has launched an agent (actually hardly anyone has...) but who controls the profile, the permissions, the context and the memory that makes the agent useful.
Because without that, the demo may look clever..... but the trip can (and will) still go wrong. How you clean that up is equally important when its trust that you're actually selling, especially if you want the next booking also.
Lots of sharp comments on the original post.
Radisson gets a ChatGPT app - Accenture makes bank
I don’t know anyone who thinks that “apps” within ChatGPT are going to work? I don’t know anyone who actually uses the @ to call them. This isn’t Slack.
Yet onward we go building them.
This article starts as a story about Radisson but ends with a financial analysis of Accenture who manage the process of getting Radisson onboarded.
The article suggests: “The partnership showcases how Accenture is applying its generative AI and data capabilities to enable "agentic commerce,"“. OK. Is that what a GPT app is? Not sure about that.
Anyhoo - apparently Accenture’s financial outlook is good.
The dream of agentic commerce
Gilad Berenstein laid out what he is hoping for as we move into the next cycle - agentic commerce:
"Agentic Commerce is about a lot more than faster product research/discovery or deeper personalization. Agentic Commerce is fundamentally a reinvention of how we create, sell, and capture value. An opportunity for a true Win Win Win revolution. There are many steps between where we are and the promised land, but we can't get there without some imagination and vision."
"So here are a few things I am dreaming of...
Agentic Negotiations Lead to Win Win Win outcomes
Dynamic Offerings & Experiences
Enhancement of Essential Operations
Restaurant Reservation & Revenue Optimization
Productizing Risk & Better Alternatives
Family & Group Trip Reconciliation"
I’m definitely here for the negotiations. Not sure the incumbents are equally as excited about ‘everyone’ winning. I guess that is where disruption comes from though!
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:
Marketplace Spotlight: TravelAI
TravelAI buying the Sonder brand and relaunching Sonder.com as an AI travel site with curated rentals caught my eye because this is bigger than a feature story. It is a distribution story.
Instead of trying to build trust for an unknown AI brand from scratch (the normal way TravelAI operates) they have bought an existing travel name with consumer recognition and pointed it at a new AI-led model. That is a pretty blunt signal that in travel, brand, intent and traffic may still matter as much as the underlying model. Here they are starting from more than halfway. Sonder sold rooms. TravelAI sells rooms. This isn't square pegs and round holes.
Are we are going to see more of this?
In Australia the old Ansett brand was revived similarly a few months back. It too garnered a ton of press. I haven't heard of it since. The new owner bought off-the-shelf "Travel Club" software and bolted the two together. This move by TravelAI isn't that. They already have the skills to build brands. They know the SEO game. They are good at ads.
We will see AI-native players buying, licensing or rebuilding around known travel surfaces because the hard part is not only generating an itinerary. It is getting the customer, earning trust, and turning that intent into something bookable. Mauricio Prieto covered this off in his newsletter this week very well - go check that out.
If you have an AI business in Travel and looking for people to notice you, you can sign up to the marketplace for peanuts (top right corner, 5 mins, bring your logo).
I’ve priced for bootstrapped startups but also accepting larger companies too.
Got a tip or seen a story I’ve missed? Let me know by simply replying to this newsletter.
Can AI see you?
An important PSA this week from Janette Roush about making sure your website is visible to the AI bot (overlords).
I got to hear Stephen Burns speak last week at Bonafide’s Align:AI conference about the three components of making your website visible to AI crawlers and found this roadmap to be instructive:
1️⃣ robots.txt Directives: Configure to explicitly allow AI crawlers (like CCBot and GPTBot) rather than disallowing them by default.
2️⃣ CDN & Edge Settings: Adjust firewalls and Cloudflare settings so AI bots aren't quietly blocked at the network edge before reaching your site.
3️⃣ Server-Side Rendering: Deliver pre-rendered HTML instead of relying on client-side JavaScript that AI crawlers can't execute.”
Burns then jumped in comments with a tool you can use to check these things in seconds. No sign up required. Totally FREE.
Slack Group!
The Slack group is full of the brightest minds in ai in travel.
This week there was a break down of the new Opus 5 model.
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 first podcast for H2 will be with you early next week. Have you missed the dulce tones? 😆
If you have an AI story happening at your company you think the travel world should know about - let me know and lets get you on.
Partner with Us
Trying to get something sourced or built and not sure where to start or looking for an objective opinion (we’ve built systems for retail travel agencies, agentic voucher to ticket solutions and lots more); or |
Looking for exposure to a travel audience of C-suite decision makers for your AI solution (see marketplace above); or |
Looking for someone to speak at your conference on AI in Travel, I’ve recently been a part of the Tourism Australia series on AI and will be speaking at the Arival conference happening in Brisbane in late June. |
- please fill in this brief form (30 seconds) |
Most clicked last week was the link to the FORA funding.
That’s it - you’ve made it to the end of this edition. If you’re thankful for this newsletter - you can always buy me a coffee.
I’ll be putting the result of the most clicked post in next week’s edition so you can see where others are focusing. If I’ve missed something, you’ve got a tip or any feedback at all - you can simply reply to this email and it will come straight to me. I’m doing this for You so please don’t be shy to tell me what you think
Artificial Intelligence (AI) Artificial intelligence leverages computers and machines to mimic the problem-solving and decision-making capabilities of the human mind. (source IBM)
Generative AI (GAI) is a type of AI powered by machine learning (ML) models that are trained on vast amounts of data and are used to produce new content, such as photos, text, code, images, and 3D renderings. (Source Amazon)
Large Language Model (LLM) is a specialized type of artificial intelligence (AI) that has been trained on vast amounts of text to understand existing content and generate original content.
ChatGPT - Open AI’s LLM; sometimes referred to by its series number GPT3; GPT3.5 or GPT4. These are used by Microsoft & Bing.
Gemini - Google’s suite of LLM.
If wanting to go even deeper into the AI lexicon - check out this handy guide created by Peter Syme for the tours & activity sector
