What the AI-bots read this season is what AI assistants tell your new guests next season

A venue music profile is a machine-readable record of what the music in a room sounds like: tempo, energy, acoustic or electronic, vocal or instrumental, and whether the volume holds steady through the evening or climbs later. It is published at a permanent web address in a format that AI systems read directly, not one people browse.

When someone asks an AI assistant where to eat on Friday and describes the type of restaurant or hotel they want, the AI assistant needs a source that describes the atmosphere. AI assistants request verified records and answer using the data they can check.

No page opens, and no click occurs between the question and the recommendation.

PlaceProfile has published 5,227,098 of these records, covering venues in 193 countries. Any AI assistant can read each one today. Claiming one is free and requires no account, no sales call, and no demonstration. Most owners still haven't read their PlaceProfile. They are still buying clicks, even though the corpus that would answer for them is already built and paid for.

What people ask for

The Gemini app reached one billion monthly active users in August, with 63% of users speaking to the AI assistant rather than typing, and more than 100 million active users on iOS alone. One in five Gemini Live interactions now includes camera or screen sharing, so the request can carry the room itself. ChatGPT serves more than a billion people every week. Gemini is built into Android and Workspace, not a page anyone visits.

Across nineteen surveyed markets, between 48% and 89% of adults already begin their information queries with an AI assistant. For a specific question, such as choosing a restaurant on a Friday, 41% of urban Indian searchers, 30% in Germany, 23% in Australia, and 16% in the United States start with an AI assistant. In urban India, 60% start at least one AI assistant query every day.

People say more when speaking than when typing. A typed query runs to three or four words. A spoken request is the full brief: quiet enough to talk in, but livens up on a Friday for after work, and not playing the same playlist as every other place, I want some old smooth jazz.

Music is among the first things people name. Until recently, an AI assistant had no source for it, so it answered with a category and a star rating. A category establishes that the place is an Italian restaurant. It says nothing about whether the restaurant plays soft acoustic sets through dinner or raises the volume after nine. Two venues with identical listings can sound nothing alike, and that difference is what the request was about.

The music profile

PlaceProfile is the venue side of ProofProfile, applying the same method to physical rooms.

Every listed venue already has a free profile, derived from a category-matched sample of 40 recordings and labeled as a category archetype. A paying venue replaces the sample with its actual playlist, then recomputes the profile and relabels it as a venue playlist. A typical profile shows what venues of that kind usually sound like. A venue playlist shows what this room actually plays.

Both share the same address, so an upgrade changes only what the profile says. An AI assistant can search by location, category, and musical atmosphere, open a single venue, read the recordings behind its summary, and recompute that summary itself.

Where the evidence comes from

The profiles draw on the ProofProfile registry: 2,006,172 artists and 9,900,355 recordings, each with a CopyrightChains registration and a confirmed Bitcoin timestamp, achieving 100% verification coverage. The material comes from the major open music databases as they existed before 2022, and the corpus was sealed in August 2026.

A bot reading the registry isn't asked to take anyone's word for anything. It reads the record, computes the fingerprint, and compares the result with the timestamp Bitcoin holds, all in under a second, with no help from us and no way for ProofProfile to influence the outcome. The registry opened to all AI bots in July. The first requests arrived within a minute, and many bots asked for proofs rather than only the records, at the volumes the published dashboard reports as floors alongside its refusal counts. Identifiable AI crawler traffic rose from roughly 8% to 13% of measured web sessions over the past twelve months.

The timestamp establishes that a record existed in a specific form at a specific time. It does not establish who made the recording, who owns it, or when it was created. Applied to a venue, the atmosphere claim ceases to be an adjective supplied by the owner and becomes arithmetic over a named set of recordings, each checkable by whoever is reading.

From recommendation to booking

Gemini can already perform actions across more than forty popular Android applications. Meta launched Muse on September 8, an AI assistant that integrates with external applications to handle tasks such as travel booking and payments. A guest can describe the evening they want and leave with a table reserved, all within a single conversation.

The venue that gets named in that conversation is the one whose record matched the description.

What is ready to claim

Paid placement is priced per click and stops when the budget is exhausted. A music profile is read when the recommendation is made, then again on the next question, and on every subsequent question.

Google Search advertising grew 16.8% year over year, with total Google advertising up 14.4%. Amazon's advertising revenue reached approximately $19.8 billion, up 26%. ChatGPT Ads reached a $1 billion annualized run rate in under 200 days. The 100 largest media publishers spent an estimated $113 million on paid search in July alone, 41% more than a year earlier and 274% more than three years ago.

On September 10, Amazon began placing selected advertisers inside ChatGPT through Amazon DSP on CPC or CPM terms. A company that does not own the AI assistant now earns from within the answer. A venue with a published music record holds a similar position.

AI-generated answers appeared in 86% of Google's People Also Ask results in August and about 97% by early September, across roughly 19.2 million English-language queries, up from around 12% fourteen months earlier. Questions are increasingly resolved where they are asked, using records the system can check.

Publishers that adapted early are now stable. People Inc. (publisher of People, Food and Wine, Travel and Leisure, and Southern Living) reduced its dependence on Google from over 70% of its audience to 20%, kept revenue roughly flat despite a fall of more than a fifth in sessions, and shifted over 40% of its business to revenue that never required a webpage visit, including direct licensing to OpenAI, Microsoft, and Meta.

The distinction that made that work separates training value, collected once when a model is built, from inference value, which an AI assistant needs at the moment it answers. A venue's music record is inference value. It is what an AI assistant requests when someone asks, out loud, for a place with the right sound on a Thursday night.

What AI bots read this autumn determines what AI assistants say next year about where to go and how it sounds when you arrive. A venue whose music record can be verified is described in its own terms, using the playlist it actually plays, and every record already includes the venue's name.

Replacing the sample with the real playlist takes an afternoon. Venues that do it this autumn are the ones AI assistants will be describing accurately next year.