Skip to content
HotelScaleUp
News, data, and strategy for independent hoteliers.
Technology
AI search, brand accuracy, hotel marketing, structured data, Eurostars Langford, original research, guest-facing content

Three of four AI assistants told us a Miami hotel was the former Miami National Bank. The city's own designation report says otherwise.

Three of four AI assistants described Eurostars Langford as the former Miami National Bank in Beaux-Arts style. Miami's 1988 designation report names the building City National Bank and the style Neo-Classical, and the original 1926 tenant was Miami Bank and Trust. All three assistants were repeating the hotel's own marketing copy. One assistant also advertised a rooftop bar that has been shut.

By Guneet Lamba
 AI assistants repeated a hotel's own incorrect history
Three of four got the building wrong the same way. The source of the error was the hotel's own copy.

Asked to describe Eurostars Langford in Downtown Miami, three of four AI assistants said the building was formerly the Miami National Bank, or the National Bank of Miami, and described it as Beaux-Arts. The City of Miami designation report of 17 November 1988 registers it as the City National Bank Building and identifies the style as Neo-Classical.

This piece is for independent hoteliers who want to know what determines the accuracy of what AI assistants tell guests about their property. In this case the determining input was not the assistants. It was the hotel's own published description.

The assistants reproduced the property's marketing copy, including the parts that are wrong

The building at 121 SE 1st Street was designed by Hampton and Ehmann, built in 1925 and opened in spring 1926 for Miami Bank and Trust. City National Bank and Trust acquired it in June 1926. Florida National Bank took the assets after City National went into receivership in December 1930. The Langford name came later, from the contractor George Langford. Stambul USA bought the building in 2012 and opened it as a 126-room hotel in 2016.

The hotel's own listing describes the property as formerly the Miami National Bank, in Beaux-Arts style. Both claims are wrong, and both came back to us from three separate assistants in near-identical wording.

Comparison grid of AI hotel descriptions from four assistants showing three repeated the wrong bank name and architectural style, and one wrongly listed the rooftop bar as open
Answers to a direct question about Eurostars Langford, Miami. Pink cells are wrong or outdated. ources: HotelScaleUp test, 9 September 2026; City of Miami historic designation report, 17 November 1988.

Only one assistant flagged the problem. Claude ran a live search, found the conflict and said so, noting that sources disagree on whether the original tenant was Miami Bank and Trust or Miami National Bank and on 1925 versus 1926, and advising that the marketing copy be treated loosely.

Omissions travel the same way as errors

Louis B. Mayer, the film producer, owned the building from 1952 until his death in 1959. It is documented, checkable and absent from all four answers.

None of the assistants mentioned it, because the hotel does not. What the assistants returned instead was the property's preferred history: the vintage phones, the steamer-trunk luggage stands, the original elevator control panel. Gemini added an original lobby mail chute, gear-like dials above the elevators and a second-floor library, none of which the other three mentioned.

The pattern in both directions is the same. The assistants are summarising the corpus of text that exists about a property, and for a small hotel the largest single contributor to that corpus is the hotel. That is a different lever from the one that decides which review snippets a platform surfaces first, and it is one the property actually controls.

Two assistants each gave half the heritage story, and neither was wrong

On the designation question, the split was not error but partial retrieval.

Claude reported the City of Miami local designation of 17 November 1988. The fourth assistant reported the National Register of Historic Places listing of 4 January 1989. Both happened. The building carries a local designation and a federal listing, granted seven weeks apart, and each assistant surfaced one and not the other.

A guest reading either answer would come away with an accurate but incomplete account, and no signal that anything was missing.

Gemini sold a rooftop bar that has been closed

The clearest commercial error was an amenity.

Claude reported that the rooftop became Bloom Skybar in 2018, that Yelp lists it as closed, and that guest reports describe scaffolding, and it warned that booking sites still advertise the venue. Copilot said the hotel's official site lists the rooftop as temporarily closed and advised against booking for the rooftop. The fourth assistant said the rooftop is temporarily closed for improvements.

Gemini described a chic rooftop bar as a highlight, twice. Its written answer listed the bar among the property's attractions with no mention of closure, and its attached Google Hotels module described the property as a Beaux-Arts hotel with an upscale restaurant and a rooftop bar.

Three assistants had the closure. One was still selling the amenity, including inside a Google-sourced product panel. The Langford holds a 4.1 rating across roughly 1,700 Google reviews, and the recurring complaints in that rating are closed amenities and rates priced as though the rooftop were open.

The competitor-redirection problem did not reproduce

Trade coverage has warned that assistants asked about one named property will push a guest toward rivals. In our brand-name query, that mostly did not happen.

Claude, Copilot and Gemini all answered the third query about the property itself and named no competitors. Only the fourth assistant offered a comparison, and it framed it as an offer rather than a redirect, naming RS Boutique Suites, YOTEL Miami and The Grayson as properties it could compare against on request.

Copilot did something closer to the warned-about behaviour one turn earlier, when it recommended the Langford as a genuinely independent stay and then, asked for detail, told us the property is operated by Eurostars Hotel Company and is not independently owned in the way a family-run property would be. The recommendation and the correction came from the same assistant two messages apart.

Where the assistants did volunteer competitors, it was against the property's ownership status rather than its rooms or its rate. That classification is being made about hotels rather than by them, and nothing in any of the four answers explained how it was reached.

Guneet Lamba
Written by
Guneet Lamba
Content Marketer

Guneet Lamba does content and SEO at PriceLabs, where she writes about dynamic pricing, revenue management, and how operators actually run their portfolios. Her work appears across the PriceLabs blog and Rental Scale-Up.

Related stories

 TikTok did not make them. Creators did, for commission, on a property that cannot be booked there.
AI search, guest-facing content, hotel marketing, brand accuracy, Expedia Group, Booking.com, original research

Seventeen AI videos sold suites a Sedona resort does not have. TikTok did not make them, and the resort was not bookable on TikTok at all.

Skift ran videos of the Ambiente Sedona through a detection tool and found 17 flagged as AI-generated at 80% confidence or higher. They showed an atrium suite that does not exist and a pond that is not there, and one sold the 40-room adults-only resort as family-friendly. One had 315,000 views. The videos carried booking commission tags. The property cannot be booked on TikTok.

By Guneet Lamba · Sep 14
We ran the same query four times. The lists barely overlapped, and that changes what optimising means.
AI search, AI recommendations, hotel discovery, Claude, Gemini, Copilot, original research, Miami

Four AI assistants named 30 boutique hotels in Miami. Exactly one appeared on all four lists.

Four AI assistants named 30 boutique hotels in Miami. Exactly one appeared on all four lists. On a second query about independent hotels near a named landmark, 11 properties came back and none were unanimous. The concentration operators have been warned about did not appear. Fragmentation did, and it means there is no single set of AI results to optimise for.

By Guneet Lamba · Sep 9 · 4 min read
Fully connected hotels report a 31-point revenue advantage over hotels with basic connectivity.
Technology

AI just turned your channel manager from back-office plumbing into who controls your guest.

Expedia Group's own Censuswide survey of 1,500 hotel decision-makers, published June 23, 2026, found that fully connected properties are 31 percentage points more likely to report improved occupancy, ADR, or RevPAR than hotels on basic connectivity. A separate RateHawk and Phocuswright report, published on Hospitality.today on June 24, 2026, argues that AI agents can only book inventory that's machine-readable, and that work is happening inside wholesalers and channel managers, one layer removed from the hotel itself. For independent hotels, the question is no longer whether they're connected. It's who controls what that connection can do.

By Minal Mehta · Aug 17 · 2 min read