A 85-ruble Colosseum: Stanza used AI to churn out 1,248 sightseeing guides. Busy on the charts, cold in the bank account

Stanza is a travel audio guide app from a one-person company in Bulgaria. Using AI for scripts and voiceovers, it rolled out 1,248 attractions in 15 languages in two months, and it appeared on the travel Top Grossing charts in 9 to 16 countries' App Stores every day. The charts look like it's taken off, but globally it has only one or two ratings. What's worth taking is how to build this 'content factory'. The results themselves aren't worth copying.

No. 11 in one country, one global rating

In Nov 2026 2, Stanza reached #11 on the Russian App Store travel Top Grossing chart. The same week it hit #18 in Estonia and #30 in Finland, and it also appeared intermittently on charts in Poland, Italy, Austria, Spain, and Brazil (App Store travel Top Grossing charts in each store, 10-28 to 11-06).

But it has only 1 ratings in the Russian store, rated 2 stars; the US store has none, and its store page shows 'not enough ratings' (App Store, 11-06). Its Google Play installs are 500+.

Put those two sets of numbers side by side and the conclusion is blunt: this chart is easy to get onto. In Russia's Travel Top Grossing top 40 there are small utilities like bird-call and animal-call apps, and in the top 25 there's a currency converter. Stanza sells one guide in the Russian store for 85 rubles, about 1 USD. One or two buyers a day is enough to crack the top 50.

The rank trajectory tells the same story. Finland was 11-02 at #30, then 47, 56, 65, 74, dropping about 9 spots a day, and Estonia fell from 18 to 41 and then off the chart. A curve like that means someone bought one that day and nobody bought after. So "charted in a dozen countries at once" just reflects how many sights and languages it covers. Somewhere in the world, someone pays $1 every day. Judging from the rank pattern, its global paid orders per day are in the single digits (inferred).

What it sells

Open the Colosseum guide in Rome and it has 29 stops. Stop one starts with the Arch of Constantine: "The structure in front of you is the Arch of Constantine, completed in 315 AD…" Stop three explains that the name "Colosseum" actually comes from a 30-meter bronze statue of Nero nearby. Each stop has a two-to-three-hundred-word script with a CC-licensed photo from Wikimedia Commons, and the attribution is printed under the image. The whole set is AI-written scripts with AI-synthesized narration (inferred, with high confidence, from the writing style, the image sourcing, and the pace of new releases).

The store listing catalog runs from the Giza pyramids and the Taj Mahal to the Acropolis, Chichen Itza, the Alhambra, and Angkor Wat. The official site says "1,248 guides, 110 countries, 15 languages," and Italy alone has 130 guides across 49 cities. The features are standard for this kind of app: GPS-triggered vibration alerts when you reach an outdoor landmark, offline download for listening, and a "digital passport" for exhibit recognition and stamp collecting.

Pricing has three tiers. Each guide offers the first two stops free, a single guide unlocks for $0.99–4.49, a whole-country pass costs $8.99–11.99, and there is an annual subscription (US App Store in-app purchase list, 11-06; I didn't see the annual price in the top rows).

Where users come from: it turns each landmark into three things

No trace of it in paid ads: Meta and TikTok's public ad libraries show nothing. Nobody talks about it on TikTok, X, or Reddit. Its traffic comes from three places, and all three run on the same system.

One landmark = one web page. On the stanzapp.com sitemap, each language has 1,485 URLs, and across 15 languages that adds up to roughly 22k landing pages, all under the path "/guides/country/city/landmark". The Colosseum page publishes the full script for the first few stops, with a button underneath: "Download the free Stanza App." Some URLs contain garbled slugs (for example nikk-t-sh-g and brazil/unknown), which look like output from a batch script. People who search long-tail terms like "Colosseum audio guide" land on these pages.

One landmark = one in-app purchase = one store keyword. It has over a thousand in-app purchases, each named after a landmark, such as "Pompeii Archaeological Park Guide." App Store search indexes these names. The store title is translated separately for each storefront: "Аудиогиды для поездок" in Russian, "Audioguide di viaggio" in Italian, and "audioguides de voyage" in French.

One update = one store event. It announces new releases through App Store In-App Events, such as "PREMIERE: Machu Picchu & 548 New Guides." These events can show up in search results and on the Today tab.

The website also has a partnership page that tells museums "zero cost, remote production, no site visit needed" and offers to produce their multilingual guides, aiming for the official-recommendation route. Whether any deals have closed isn't visible from outside.

Who it's competing with: this track is old and crowded

(App Store storefronts, 11-06)

There are two more layers below. The first is the old-school model of paid one-app-per-landmark guides from the past decade or more, such as official or semi-official guides for the Vatican Museums and the Alhambra. They have few ratings but stay on category charts year-round. The second is a batch of AI-guide apps that appeared in the past 2026 years, including Colosseum Tessera, WegSaga, Waltru, RoamWithMe, and Street Whisperer. Most cover a single city or landmark, with ratings from 0 to 8. This batch emerged at the same time as Stanza, and they all use the same "LLM script + speech synthesis" approach, so none of them is copying the others. Above them is the ceiling: GetYourGuide and Tripadvisor sell audio-guide tickets, and Google Maps and Gemini can answer questions about landmarks directly.

Stanza has one advantage: scale. One app, over a thousand landmarks, and 15 languages, including Russian, Polish, Arabic, Hindi, and Hebrew, which few others serve. But even with that scale, it doesn't stand out from its peers. There are no human voices, no brand, and no accumulated reviews. The 1 2-star ratings in the store are its entire trust signal.

If it were you

Grade C: worth learning from, not worth copying. The playbook is clear, cheap, and anyone can set it up, but Stanza spent 1,248 landmarks to get single-digit daily orders at $1. The track has a decade of older players and a batch of same-era AI newcomers, with big-tech free features pressing down on it.

Who it suits: People who can do programmatic SEO and multilingual ASO, and want a side project where content costs almost nothing and grows slowly through long-tail search. Or people who already run travel content sites or landmark traffic and want guides as a monetization exit.

Who it doesn't suit: People who expect clear revenue within a few months. People who can't keep AI-written historical facts accurate, since the most common complaints about this kind of product are historical errors and synthetic voices that wear you out. And people who can build apps but not earn search traffic.

How to do it: Stop trying to cover "every landmark in the world." Someone already has, and the wider you spread, the thinner you get. Pick an angle it does poorly and go deep: one country or one language audience (for example, Chinese, Korean, or Arabic-speaking independent travelers in Europe), or one type of scenario (museum trips with kids, road routes, religious pilgrimage sites). Make the content accurate, the voice good, and build up reviews. Copy the structure directly: one landmark maps to one web page, one in-app purchase, and one store keyword, with the first two stops free and a country pass to lift order value.

How to get users: Search is the main battlefield, covering both Google long-tail queries and App Store search. Next come partnerships with landmarks and museums (it has asked on its website, though I haven't seen results), plus travel bloggers and "how to see X" content on Xiaohongshu and TikTok. Ship updates one to two months before peak travel season, and pair them with store events.

When to start: Before European peak season, meaning before April 2027 3. The catalog and pages need to be indexed in search by then. Google could crack down on bulk AI pages at any time, and the later you start, the less stable this traffic will be.

What would show I'm wrong: ① If Stanza's ratings climb to several hundred next summer and it starts appearing on overall charts, that would show "mass AI pages + SEO" really does work, just slowly, and I'm underrating it. ② Conversely, if its landing pages largely disappear from Google search, or most similar apps still have single-digit ratings a year from now, then this path doesn't even capture long-tail traffic. Stay away.