When someone asks an AI assistant for a quiet cafe or a lively bar, something has to tell it what your venue actually feels like. That is what we publish, and every number in it can be checked.
Verify it yourself · Why it matters · What it does not prove
Why you can believe it
Most companies ask you to trust their numbers. Ours are built so that you, or the AI reading your venue's page, can check them.
A venue's atmosphere score is simply the average of the forty songs listed on the same page. Add them up yourself and you get our number.
Twelve things, measured rather than guessed: tempo, energy, how upbeat or melancholy it sounds, how acoustic it is, how danceable, how much speech it contains, whether it is instrumental, whether it was recorded live, loudness, key, major or minor, and time signature. The venue summary is arithmetic over those numbers and nothing else, so a summary that disagreed with the songs beneath it would be visible on the same page.
Every song is timestamped in a public record kept outside PlaceProfile, so if we altered anything after the fact, it would show.
Each song's record is fingerprinted, and that fingerprint is written into the Bitcoin blockchain through OpenTimestamps. Anyone can recompute the fingerprint from the record we publish and confirm it against the block. We harvest these records verbatim and never regenerate them ourselves, which is what makes the check worth running.
Every song links out to its own entry in an independent music registry. None of the checks route back through us.
The catalogue is drawn from datasets that predate generative AI, so these are verifiably real recordings by real musicians rather than synthetic filler. Each entry keeps the registry's own labels, so following a link from a venue page to the registry needs no translation.
Each venue page publishes the same data in a structured form that machines read directly, linked from the page itself. Every figure described above appears there alongside the evidence it is derived from, so the checks can be run without contacting us.
Example venue: Hotel Oceania Rapa Nui, Isla de Pascua.
Why it matters
The web knows what every venue is. It has almost nothing on what a venue feels like, which is exactly what people ask for.
AI assistants increasingly stand between customers and venues, and the contest is no longer about ranking in a list of links, it is about being the source the assistant relies on. To answer "a quiet cafe with warm acoustic music near me," an assistant needs atmosphere as data, and no such data exists at scale. Venues generate that signal every day in the music they play, and never publish it. We turn it into measurements and publish it at a permanent address.
The data itself is the thing AI reads, so there is no content calendar to keep and no SEO arms race to run.
Each song is measured across twelve dimensions, and the venue's profile is the average of them. Until an owner connects their own music, a venue carries a sample matched to what places of its kind typically sound like, and the page says so plainly. Connect a real playlist and the same page becomes a measurement of that specific venue.
Am I findable to AI? Your dashboard answers it from your own visits, and is honest about what it cannot tell you.
You see which AI services read your page, and whether each one came because a person asked it something or because it was collecting pages in the background. Anything can claim to be ChatGPT, so we mark a visitor as confirmed only when we can independently check where it came from. The counts are actual visits, never padded.
What it does not prove
When an AI service reads your page, that tells you it read your page. It does not tell you what happened next.
Nobody can show you that a page was remembered, or chosen, or quoted back to a customer, because those steps happen inside systems none of us can see into. We show you the parts that can be measured and label the rest as unknowable, rather than dressing a page view up as a booking.
No honest global count exists, so any figure would be invented. We tell you how many AI readers we recognise by name, currently 119, and we call that what it is: our list, not a census of the internet. Inventing a number would be the first unverifiable claim on a page whose whole argument is that our numbers can be checked.
Live, real data
Capannone is a real restaurant in Stockholm. Whatever the dashboard shows is what readers actually did, including the days nothing happened.
Open the live dashboard →public, no sign-in · More live examples →