Travel Tech Essentialist #195: Essence
When AI makes execution cheap and fast, value migrates to what machines can't replicate: the judgment to know what problems matter, the perception to spot when reality shifts before others do, and the lived experiences that reveal friction no dashboard ever will. This edition looks at what becomes precious when everything else becomes abundant…from domain expertise to physical spaces to the moments that actually annoy you enough to fix them.
1. The new speed of software
Meta spent $2B to acquire Manus, an AI agent startup. In only 10 days, Anthropic built and shipped their competitor, Claude Code, a full agent-powered dev tool, with every line of implementation code written by AI.
The big story is how fast software is changing, collapsing the layer between intent and execution. Shashwata Bhattacharjee’s post shows how engineers’ roles are shifting from writing code to orchestrating AI agents, managing validation pipelines, and identifying failure modes. The result is roughly 95% cost savings, 92% timeline compression, and development capacity that scales exponentially with human oversight instead of linearly with headcount.
In travel, agent-based planning tools, dynamic packaging, pricing systems, AI-native customer service and internal tooling from fraud to schedule optimization are all ripe for this kind of acceleration.
2. Management beats coding now?
That same compression is happening at every level. Ethan Mollick ran an experiment at Penn in which he gave executive MBA students four days to build a startup from scratch. Most had never written code. They used Claude Code and AI tools to build working prototypes, run market research, and create financial models, etc… What normally takes a full semester took four days. The prototypes had fully working features, the market analysis was sharp and the financial models made sense.
These weren’t CS students. They were doctors, managers and executives taking night classes. What they did bring to the table was years of experience scoping problems, defining deliverables, and spotting when a financial model or customer analysis was off. The students who performed best were those who already understood how to delegate and set clear goals. Their real-world experience gave them mental frameworks that translated directly into effective AI prompts.
Mollick built an equation for when handing work to AI makes sense. It has three components:
Human baseline time: how long would the task take you to do yourself
Probability of success: how likely the AI is to produce output that meets your bar on each attempt
AI process time: how long each round takes (prompting, waiting, evaluating)
If something would take you 10 hours but you can get AI to nail it in 2 hours of back-and-forth, you saved 8 hours. If something would take you 1 hour but you spend 2 hours getting AI to do it right, you wasted time. A key variable is Probability of Success. That’s where management skills matter. People who can write clear specs, evaluate output quickly, and give useful feedback get better results in fewer attempts. The AI doesn’t magically know what good looks like. You have to tell it. Read + Ethan Mollick
3. Your travel domain expertise matters more now
Mollick’s experiment has direct implications for travel founders. You can build a booking flow prototype in hours instead of months. Test pricing strategies in minutes. Generate customer service responses at scale. The constraint is whether you can explain what good looks like.
Can you describe what makes a trip planning experience work? What should a helpful rebooking email say? Which checkout friction actually costs conversions? Years in travel operations, customer service, or product work gave you judgment about these things. That judgment is what AI needs from you. The “soft skills” companies used to treat as nice-to-haves are now essential. When everyone can code via AI, the advantage goes to those who understand travel problems well enough to direct AI toward real solutions.
Your competitor who’s never worked in travel can spin up a booking platform over a weekend now. What they can’t do is recognize that handling rebooking chaos when connections get missed might be more of an issue than finding flights. They don’t know which support questions signal bugs versus confused users working around bad UX. You know these things. That knowledge just got more valuable.
Can you articulate what good service looks like clearly enough that someone else could execute it? If you can’t explain it to a person, you can’t explain it to an AI either. Domain expertise always mattered in travel. Now that execution speed is no longer the bottleneck, the advantage goes to those who know what to build.
4. ChatGPT’s 4% agentic commerce fee just went live
Starting January 26, Shopify merchants using ChatGPT to sell will pay OpenAI a 4% fee on every transaction. That’s on top of standard payment processing costs, bringing the total merchant costs closer to $7.10+ on a $100 order vs. $3.10 for traditional e-commerce.
As Dwayne Gefferie points out, this puts ChatGPT in direct contrast with Google Gemini and Microsoft Copilot, which currently charge 0%. OpenAI is betting their 800 million weekly users justify premium pricing. OpenAI controls distribution and now it’s extracting margin from it. They’re positioning closer to Amazon’s marketplace fees (8-15%) than affiliate rates.
5. AI’s balance sheet reality check
ChatGPT still dominates public perception of AI, with 800M+ weekly users and the #1 spot in Apple’s App Store. But OpenAI is burning cash fast at an estimated $14 billion in 2026 alone. Between the new 4% Shopify fee and ChatGPT ads, OpenAI is in a rush to monetize. For a company that once positioned itself as a research lab, this is a clear shift.
Meanwhile, Anthropic is pacing itself differently. Claude Code hit a $1B run rate in just six months. The company is targeting $20B–$26B in ARR for 2026, with 80% coming from enterprise customers, and expects to be cash flow positive by 2027 (source).
Google isn’t rushing at all. With $300B in annual ad revenue, Google's 650M monthly Gemini users won’t be seeing ads anytime soon (source).
The race is still open.
6. Tryp.com vs eDreams
eDreams has sued Tryp.com, accusing it of abusing the eDreams Prime subscription system to resell discounted Prime fares to Tryp customers and auto‑canceling Prime trials via AI. eDreams then canceled hundreds of tickets, which Tryp says included non-Prime fares. Tryp argues it’s merely acting as an OTA, representing users in finding the lowest fares, while eDreams similarly “scrapes” Ryanair. Airlines license inventory to intermediaries via explicit contracts and APIs, but Ryanair and eDreams have no such agreement, which makes eDreams’ position toward Tryp.com look uncomfortably similar to the airline stance it has fought for years. The clash has become a test case about fare access, dark‑pattern subscriptions, and whether an OTA can block a rival from using its consumer‑facing fares while still claiming a right to “open” access to airline inventory.
Maybe Tryp should spin out a new business to help people cancel their eDreams Prime subscriptions…😉
Quick poll…
7. The problem with most problems
Mike Maples famously passed on investing in Airbnb, and in this piece, he explains why. The problem wasn’t with the idea or the founders. It was with his framing.
Most founders look for pain points to solve. Ask someone in 2007 what frustrated them about travel accommodations. They’d say hotels were too expensive or poorly located. Nobody said, “I wish I could stay in a stranger’s apartment.” Those are conventional problems. You’re asking customers what they want improved within the current rules.
Maples argues the biggest opportunities come from explanatory problems: moments when reality stops matching what everyone believes should be true. The Airbnb founders launched a basic WordPress site offering air mattresses to make rent. The response surprised them. Standard theory said strangers wouldn’t trust each other with accommodations, but people did anyway. The anomaly exposed a broken assumption.
Most investors dismissed the anomaly with excuses: “Those early customers are a tiny market.” “That only works in San Francisco.” Each excuse let them ignore the evidence and stick with the old explanation.
In an AI world where execution is cheap and fast, solving conventional problems gets commoditized. What’s scarce now is perception…the ability to spot when reality has shifted and build a new explanation before anyone else sees it.
8. The rainy night every product needs
Jim Moylan died last month. You’ve never heard of him. But you’ve seen his invention thousands of times. In 1986, the Ford engineer pulled into a gas station on a rainy night and parked on the wrong side of the pump. He got soaked repositioning the car, then drove back to the office, still dripping, and typed a memo before taking off his coat. His idea was to put a small arrow next to the fuel gauge showing which side the gas tank is on. Ford added it within months, and then every automaker copied it.
Moylan’s frustration was real because he was living the customer experience. The arrow works because he felt the problem firsthand and was annoyed at the wasted time, just like countless drivers before him who couldn’t do much about it.
Travel companies that win obsess over tiny friction points. The ones that lose treat customer experience like a dashboard metric instead of something they’ve lived through. When Ryanair CEO Michael O’Leary flies his own airline, he experiences what budget travelers experience. When Airbnb’s Brian Chesky stays in listings, he sees what hosts and guests see. When Lyft’s CEO, David Risher, spends hours each quarter driving riders, he sees real driver UX, pricing, and support issues firsthand.
The next time you design a booking flow, walk through it on mobile with a bad connection. Check into your own hotel exhausted at midnight. Call your customer service line with a real problem. Your best product ideas will come from the moments that annoy you enough to fix them.
9. The next Golden Era of travel
This new Google × Alvarez & Marsal study is a 90‑page report packed with travel data to model travel’s long‑term future. It builds predictive models on 20‑plus variables, billions of Google searches, and 90,000‑plus tourism datapoints (plus airline data) to estimate future tourism flows and outline what a ‘next golden era of travel’ might look like. They expect global travel to roughly double to about 3.5 billion international trips and around 6 trillion dollars in international spend by 2050, with APAC becoming the largest source market, Europe staying the top destination, and domestic travel remaining the industry’s backbone at over 90% of trips. They also argue that volume alone won’t deliver profitability and that AI should be the industry’s primary enabler. This well‑structured scenario set is worth a read.
10. The physical world as competitive advantage
We spend a lot of time talking about digital tools, AI agents, and automation. This piece by my friend Isaac French is a reminder of what they can’t replace. His thesis is that AI will commoditize tools, creativity, and information. What it can’t commoditize is physical beauty and human care. You can’t download an environment. In the age of infinite outputs, what people are starving for is not more to do, but places to be. Places to belong. The hunger for real environments, for truth you can touch, will only grow.
This matters for travel. When AI handles booking, discovery, and personalization, the physical experience becomes your only real moat. Not your app. Not your algorithm. The room, the light, the thousand small acts of care that make people feel something. Most operators treat place as a backdrop. French argues it’s the product, and he’s probably right.
Travel Tech Essentialist Job Board
→ Explore all 1522 open roles on the Travel Tech Essentialist Job Board now.
Airbnb | Communications Lead, U.S. Latino | $160k - $200k + Equity | USA; Remote
Booking.com | Lead, Global Communications- Channels, AI and Enablement | Amsterdam
Fora | Advisor Success Manager, Cruise | New York City | $90k - $120k + Equity
Search jobs by seniority
→ Internships (38 jobs)
→ Entry Level (293 jobs)
→ Associate (76 jobs)
→ Mid-Senior Level (434 jobs)
→ Senior (365 jobs)
→ Director / VP / CXO (127 jobs)
📩 For monthly updates on the latest roles, subscribe to the Travel Tech Jobs newsletter
Fundraising? Submit your startup to the Travel Investor Network
If you’re raising a round (pre-seed to Series D), complete your profile on our new platform. We’re partnering with Pynn, an AI-powered platform to better evaluate startups and manage deal flow. It takes about 5 minutes to complete and helps investors assess opportunities more efficiently. We review every submission. Even if we don’t follow up directly, your profile may still be discovered by relevant investors.
Travel investors (VCs, corporates, angels): If you are interested in joining the Travel Investor Network, please request an invite by completing this form (should take less than 2 minutes).
If you like Travel Tech Essentialist, please consider sharing it with your friends or colleagues. If you’re not yet subscribed, join us here:
And, as always, thanks for trusting me with your inbox.
Mauricio Prieto
I often think about how much harder it’s become to find real pain points in travel. The obvious ones have largely been smoothed over, and most AI-driven improvements can be replicated quickly by industry leaders with scale and distribution. I’ll be very interested to see how this reshapes where value is actually created in the space over the next few years, particularly on the consumer versus business side.
the Airbnb story of framing is quite interesting. Technically Jobs framed the iphone as an iPod, a phone and Internet. In the end the market decided that wasn't at all what they wanted. iPhones are communications (text mostly) devices, gaming devices, maps and other apps. The right framing mostly comes from the market. In my experience founders, owners etc - are always wrong on the framing. Once they bring it to market and get feedback we start seeing the real framing. Asking users "why do you love it?" is the key question (not what do you like).