There is no person picking the price of your hotel room. There has not been for years. The number you saw at midnight was set by software, adjusted by more software, and calculated off data you were never meant to see. That is the premise of a cracking piece by Angus Sewell, and the maddening bit is that he used to help build the machine.
In “How I Beat Airbnb’s Pricing Algorithm With Vibe Coding”, Sewell, who writes Angus the Nontechnical and spent years around hotel pricing software, argues that the professional-grade data revenue managers pay fortunes for is now buyable by the call, and that you can point AI “skills” at it to book smarter. Read the original, it earns your time.
First, meet the beast
“Dynamic pricing” means the rate moves constantly based on demand, competitor prices, how fast a given date is filling, and a hundred other signals, the same way airline seats and taxi fares do. The scale is genuinely hard to picture. Lighthouse, one of the big hotel-intelligence platforms, says it processes around 1.7 billion rates a day across millions of properties. That is the firehose your gut instinct is up against when you tell yourself “book early and I will get a deal.”
What Sewell gets right
His useful contribution is showing that “book early” is folk wisdom, nothing more. The cheapest moment to book varies by market and by how a specific date is pacing, so it has to be worked out rather than assumed. He walks through cheap data sources and a five-step workflow that writes everything into a single planning file. If you travel a lot, or you are the person in the family who always books, some of this genuinely pays off.
Where he undersells the risk
That cheap, pay-by-the-call data tier he builds on is not a stable foundation, it is a moving target. Consider the timing. On 17 July 2026, Amadeus, one of the giants of travel data, decommissioned its self-service developer portal entirely, cutting off exactly the indie developers and tinkerers this whole approach depends on. PhocusWire reported the change hits “startups, indie developers, and small OTAs” hardest. Today’s ten-cent lookup is next quarter’s “contact sales.” Build a personal workflow on these pipes and you are one pricing-page update away from a broken toy.
Then there is the headline itself. “Ten cents” is a lovely number and a slightly misleading one. The data is cheap. Your time is not. Wiring up three data providers, assembling five AI skills, and maintaining a planning file is a project, and for the once-a-year holidaymaker the maths rarely beats simply being flexible on dates. The people this actually serves are frequent travellers and small operators. That is a fine audience. It is just not “everyone.”
And there is a structural catch worth naming out loud. Consumer-side price optimisation works precisely because most people do not do it. The revenue-management systems Sewell describes are adaptive by design. If a real share of guests started gaming their booking windows with the same data, the algorithms would reprice around the new behaviour. This is a genuine edge for the individual today. It is not a loophole that survives contact with scale.
The thing actually worth being angry about
The asymmetry, and here Sewell is dead right to pull back the curtain. For years the hospitality industry has priced you using data you had no access to and mostly did not know existed. The shift he describes, where some of that data leaks out to the rest of us by the call, is a real and welcome levelling. Just go in clear-eyed. You are not beating the algorithm. You are renting a brief look at the same dashboard the house has always had.
The low-effort version
Most of the win needs no code at all. Never take the first date you check. Compare at least one channel against the hotel’s own site, Sewell found a 97-dollar gap on a single Nashville booking. And remember the headline rate is not the price until you add taxes, fees, and a cancellation policy that actually works for you. If you want the full AI-skills build, his piece is the map. For most people, those three habits capture most of the benefit.
Did you know: the cheapest listed price is regularly not the cheapest booking. Once you match room type, taxes, resort fees, and cancellation terms, the “deal” channel can end up dearer than the hotel’s own site. Compare like for like, not just the big number on the tile.
Sources
- Angus Sewell, “How I Beat Airbnb’s Pricing Algorithm With Vibe Coding”, Angus the Nontechnical.
- Lighthouse, Rate Insight (data volume across hotels and short-term rentals).
- PhocusWire, “Amadeus to shut down self-service APIs portal for developers”.