Namebeam emblemNamebeam
Get picked by AI

Who the AI assistants name: a daily public record, 8,653 rows, 26 markets, through 2026-09-09

First-party record. Published 2026-09-12 by Namebeam, a Carter Enterprise LLC company. Every number on this page was computed by script over the file it names on the date shown. Method, sample size and weaknesses printed below, zeros included.

Why this exists. Several studies this year count how often AI assistants name local businesses. All of them publish totals. None of them publish the rows. This page is the record itself: which engine, which market, which niche, which day, which names came back, with the raw answer file kept and hashed behind every row.

What was measured and how

Each day a recommendation prompt for a market and a niche is sent by API to four assistants: ChatGPT (openai), Claude (anthropic), Perplexity and Gemini. The full answer is saved as a file, hashed with SHA-256, and the strings that look like business names are extracted into a list. The same query is also read on a Google results page. A third row type, AGREE, records which names the engines agree on for a target.

The compiled file, Namebeam_E1_Dataset, held 8,653 rows on 2026-09-09, spanning 2026-08-02 to 2026-09-09 (38 distinct days). Since 2026-08-26 it grows by 282 rows a day. Row kinds: 5,752 API answers, 1,458 Google results pages, 1,443 AGREE rows. Engine rows: openai 1,414, anthropic 1,414, perplexity 1,466, gemini 1,458, google_serp 1,458.

The file carries 26 market labels. The largest by row count: US nationwide 1,483; St. Louis MO 719; Pleasanton CA 685; Kansas City MO 538; Atlanta GA, Houston TX and Dallas TX 330 each. It carries 34 niches; the largest: personal injury attorneys 660, short-term rental 597, HVAC repair 495, residential roofing 495, digital marketing agency 343, tanning salons 300. 48 targets (businesses or Namebeam's own legs) have been checked. Twelve market records built from this file are published at markets.namebeam.ai (Atlanta, Charlotte, Chicago, Dallas, Denver, Houston, Kansas City, Miami, Phoenix, Tampa, US nationwide, US), and eleven market and niche pages are published on namebeam.ai.

The record beside the count studies

StudyDate on the pageSizeWhat is public
Namebeam E1 daily record (this page)through 2026-09-09, daily8,653 rows, 26 markets, 34 nichesEvery raw answer file, hashed, in a public repository; twelve market records; eleven market pages
Local Vitals2026-04-01600 AI queries; 23 businesses on all four platformsTotals
Surfeo2026-04-08518 businesses across 4 AIsTotals
Semrush, ghost citations2026-06-09share of AI citations that carry no brand mentionTotals
Insites2026-07-2710,000 local businesses, ChatGPT and PerplexityTotals
Routeless2026-07-2630 local marketsTotals
Steady Demand2026-08-0714,000+ local AI citations, Gemini and ChatGPTTotals
Search Engine Land, Gemini citations study2026-08-19source types Gemini cites in local AI searchTotals
Bill Hartzer2026-09-09AI local business recommendations, 6 percent to 45 percent in a yearTotals

Dates are the published or modified date read from each page's own metadata on 2026-09-12. Sizes are as each study states them. Namebeam has not re-run any of them.

What a business owner can do with a dated row of its own

A row says, for one day and one engine, whether your business came back when a customer's question was asked, and which names came back instead. That is a fact with a date on it. You can show it to a partner, a lender or a client, and they can check it against the raw file. If the engines are naming a competitor and not you, the row names the competitor. If they are naming you, the row is the proof, dated, and the next day's row shows whether it held. Namebeam sells the dated receipt of your own row at $29 a month and a one-time audit at $197; the free check at check.namebeam.ai returns your first row at no cost.

Methods and limits

Of the 5,752 API answers, 2,187 named no business at all (38 percent). The name extraction is automatic and still captures strings that are not businesses (place names, words such as Phone or Certified, an occasional URL), so counts on this page are counts of extracted strings, not of verified businesses; the raw file behind each row is where a name is confirmed. One prompt per market and niche per day means a business's absence on one day is one observation, not a verdict. Namebeam's own numbers, printed as they are: 27 checks on the public proof page, 3 namings, 1 void; 156 emails to 93 businesses since July, 1 reply; 8 search impressions and 0 clicks on the proof page in the last 7 days; revenue $0, customers 0 as of 2026-09-12.

The file

Raw answer files, one per check: github.com/mrcrtr1979-droid/namebeam-visibility-index, corpus/e1. Market records: markets.namebeam.ai, twelve markets, usable rows counted per business. Free to quote with attribution: Namebeam E1 daily record, 2026, namebeam.ai/who-the-ai-assistants-name.html.

Run the free check on your own business