Questions about Xaudient

Short answers about what Xaudient does and how to check what we publish. Each answer restates a page of this site, linked below it.

What is Xaudient?

Xaudient is an AI brand visibility platform. It measures how often AI assistants name a brand when they answer the questions buyers ask, in what position and with what Tone, next to the competitors the brand has approved, and recommends what to change. Xaudient is built by Xaudient Ltd, a company registered in England and Wales.

More on this: How it works

Which AI assistants does Xaudient measure?

Four: ChatGPT, Claude, Gemini and Perplexity. Every question in a Monitoring Cycle goes to all four, and the results are shown assistant by assistant. When the assistants disagree on what to recommend, the report says so.

More on this: How it works

What does Xaudient measure?

SAV is your brand's mentions as a share of all mentions across your confirmed competitive set. WSAV (weighted share of AI voice) is the same share weighted by position in the answer and by tone. Next to them: the brand's average position in the answer, its Tone, and the same figures for each competitor in its Setup. Questions that name no brand measure its organic visibility.

More on this: Share of AI Voice · Metrics explained

What is the AI Visibility Score?

A score from 0 to 100 for how legible a brand's own website is to AI. It comes from a deterministic audit of the site, with no AI model scoring it: structured data, technical readiness and findability for AI, each scored 0–25, compared with the competitors' sites. It describes the site, not what AI assistants answer.

More on this: What we do

What is Tone?

Tone is how an AI answer speaks about a brand it names: positive, neutral or negative. English answers use a keyword indicator. Italian answers are AI-assessed where enough answers could be assessed, and marked Provisional. Otherwise Tone is reported as not measured. Tone does not change SAV, which counts mentions only; it is one of the two weights in WSAV.

More on this: Tone

How does a Monitoring Cycle work?

First comes Setup, once per market: the brand, the competitors it approves and its market. Confirming locks the Setup, and later changes are versioned and flagged in the report. Each Monitoring Cycle then asks the buyer questions, across five categories, to ChatGPT, Claude, Gemini and Perplexity, measures SAV, WSAV, position and Tone, and recommends what to change. The next Monitoring Cycle uses the same questions, competitors and method, so it shows what moved; when any of them changes, the report says so. The pace is yours: each new Monitoring Cycle adds to the pattern.

More on this: How it works

What does Xaudient look at in an AI answer?

The answer itself, one by one: which brands it names, in what position and with what Tone, and the sources it cites. That is where you see why a competitor wins a question. Answers that attribute another company's facts to the brand are flagged, and so are assistants that disagree on what to recommend.

More on this: What we measure

Which questions does Xaudient ask the AI assistants?

The questions buyers ask, across five categories: market, competitor comparisons, pain-driven, buyer journey, geography. They are checked before launch, language included. Questions that name no brand measure organic visibility. Each language and market is its own series, with its own Setup, competitors and Monitoring Cycles.

More on this: Who you are for AI

How does Xaudient relate to SEO?

Xaudient works on how AI assistants describe and recommend a brand, on the same website your search work already uses. It audits how legible the site is to AI (structured data, technical readiness and findability for AI) and says what to change on it, such as robots.txt, the sitemap and key pages served without JavaScript. It prepares AI files the brand can publish next to its pages: llms.txt, llms-full.txt, a knowledge graph and an ontology. Each Monitoring Cycle then measures whether the AI answers moved.

More on this: What we do

Who publishes the files and acts on the recommendations?

Your web team or your agency. Xaudient measures and recommends; each recommendation names one bounded action and its owner. Your team puts the files on your site and acts on the recommendations, and the next Monitoring Cycle shows what moved. Some clients work with us through their agency; the measurement is the same either way.

More on this: How it works

Does Xaudient change what AI assistants say?

No. We are a thermometer, not a thermostat: we measure what AI assistants say about a brand, and we do not control it. We never inject false claims or manipulative content to inflate a brand's visibility. What we provide are AI files and recommendations based on the measurement, and the next Monitoring Cycle measures whether anything moved.

More on this: Manifesto

What are AI files?

Machine-readable files a website publishes so AI assistants describe it accurately: llms.txt, llms-full.txt, a knowledge graph, an ontology, and disambiguation.txt, which states which other companies or names the brand is not, from confusions found in AI answers and confirmed by the brand. They sit next to robots.txt and the sitemap, and they are files, not pages of the site. Xaudient prepares them from the brand's latest completed Monitoring Cycle. Each text file starts with a provenance header: when it was generated, by what, its checksum and where to verify it.

# -----------------------------------------------------------
# llms.txt for Xaudient
# Generated: 2026-11-04T13:54:33.771955+00:00
# Generator: Xaudient AI Visibility Platform
# Checksum: sha256:04641115be63f3a0e21751f834cc07fa5cdc37119f5eb354c83b98e3e697bece
# Verify: /.well-known/xaudient-verify.json
# Crawl-Verified: true (sourced from live website audit)
# Last Crawled: 2026-11-03T11:01:15.464298+00:00
# Spec: xaudient-provenance/v1.1
# -----------------------------------------------------------
Excerpt of xaudient.com/llms.txt, as published with this page

More on this: Verifiable by design

How can I verify Xaudient's AI files?

Web content can be built to mislead the AI agents that read it, so you should not have to trust us about what is in our files. Every AI file is listed in a public manifest, /.well-known/xaudient-verify.json, with the SHA-256 checksum of the whole file exactly as served. Download a file, compute its SHA-256 and compare it with its entry in the manifest. If they match, the file is the one the manifest lists. If one byte changes, the checksum no longer matches: the file is not the one we published. A text file such as llms.txt also has a Checksum line in its header; that one covers the body only, everything after the header and the blank line that follows it, so the two values differ on purpose. A checksum shows that a file is unchanged. It does not show who published it. Our /verify page runs these checks on our own files, in your browser.

shasum -a 256 llms.txt
  "assets": {
    …
    "llms_txt": {
      "checksum": "sha256:cc439a641d8b59c1cf34cfb328a222fbec2b9bf46c89cbe309a63d04f81f7e41"
    },
    …
Excerpt of xaudient.com/.well-known/xaudient-verify.json, as published with this page

More on this: Verifiable by design · /.well-known/xaudient-verify.json

What does Xaudient sign, and how do I check a signature?

AI files are checksummed against a public manifest; recommendations and the methodology are signed (Ed25519). A signature proves more than a checksum: who issued the file, not only that it is unchanged. The methodology is our manifesto, /manifesto.md; its signature file gives the SHA-256 of the manifesto, the key_id and the signature. In a recommendation file, the signed bytes are everything before the line that starts the signature block: file_sha256 is their SHA-256, sig is the Ed25519 signature of those bytes in base64url, and key_id names the public key to check it with. The public keys are at /.well-known/xaudient-rec-keys.json, so anyone can check a signed file without asking us. AI files are not signed, and we do not call them signed: only recommendations and the methodology are.

{"family":"manifesto","manifesto_format":"manifesto-sig/v1","manifesto_sha256":"3529c00b1ea6ad1fb6dbccca6508bc1194efce56a07119cc1b97916dc0ea67f9","key_id":"xa-rec-prod-2026-10","issued_at":"2026-11-02T17:50:17Z","alg":"Ed25519","sig":"T3t39qNSpVoHNwHb5dr9HfJMJ-fWA6sr7fJAev4_caTY0t0aTeQS2KgCaiR-MY7ClQQP1rM4A-MHyuEfKEpGAQ"}
xaudient.com/manifesto.md.sig.json, as published with this page

More on this: Verifiable by design · /.well-known/xaudient-rec-keys.json

Is there an API or an MCP server?

Not yet. An API and an MCP server for AI agents are planned and coming soon; neither is public today. Some brands could run their own MCP server, so AI agents can reach accurate brand information, and Xaudient can help with that.

How does Xaudient handle personal data?

To measure a brand, Xaudient reads the brand's public website and the answers AI assistants give to buyer questions. Personal data of our users is covered by our Privacy Policy. We do not sell your personal information.

More on this: Privacy Policy