AI visibility

How AI assistants use these facts.

When someone asks ChatGPT, Claude and Perplexity about an AI hiring tool, the assistant answers from what it can read. HireAIScore publishes every score, the evidence behind it and each vendor’s reply in the forms assistants are built to read - and lets their crawlers in by name. The facts are the ones people read on the page; nothing is said only to machines, and nothing in them is for sale.

What we publish

The record, in the forms assistants read.

Every one of these restates what the human-facing pages print, from the same stored values, so an assistant quoting any of them quotes the profile.

  • 216 vendor fact sheets

    Each profile as one Markdown file at /vendors/{slug}/facts.md: score, grade, rank, every criterion with its cited evidence, the vendor's reply, documents listed on request, company facts, the conflicts disclosure and how to cite it. Linked from each profile's page head and from /llms.txt.

  • 18,445 comparison pages

    Every pair of vendors scored in the same category, at /compare/{a}-vs-{b}: the margin, each criterion side by side, and the evidence behind both. What an assistant can read when it is asked “A or B?”.

  • /llms.txt and /llms-full.txt

    A map of the site for language models, with every fact sheet listed, and the methodology, regulation explainers and analysis as one Markdown file.

  • Structured data on every page

    Product and Review markup on each profile, with the score, reviewer and review date stated exactly as the page prints them; FAQ and breadcrumb markup across the site.

How AI uses these facts

When buyers ask AI, this is the record it can read.

ChatGPT, Claude and Perplexity and other assistants answer questions about AI hiring vendors from what they can read. We publish every score, the evidence behind it and each vendor’s reply in the forms they are built to read - the same facts a person sees on the page, nothing said only to machines.

  1. 01

    A buyer asks an assistant

    A procurement lead, an HR team or a compliance officer asks about a tool, or two of them side by side.

    “Which AI interview tools publish a bias audit?” · “Vendor A or vendor B for high-volume screening?”

  2. 02

    The assistant reads the record

    Our robots.txt invites the AI crawlers in by name, /llms.txt maps the site for them, and every vendor's profile and every same-category comparison is published as text they can read whole.

  3. 03

    The answer draws on dated, sourced facts

    What it reads is the score with its rubric version and review date, the evidence behind each criterion with its source, and the vendor's own reply where one is published.

No vendor can pay to change what the record says, where it ranks, or what we publish for assistants. Assistants decide what they cite; we make sure what they read is complete, dated and sourced.

Recognised assistants

The AI crawlers we let in and count.

Each is named in our robots.txt with the site open to it. When one fetches a vendor’s profile, fact sheet or a comparison, the read is counted for that vendor. Google’s AI Overviews and Gemini, and Microsoft Copilot, read through the ordinary search crawlers, which are welcome but can’t be told apart from search, so they aren’t counted.

AI crawlers HireAIScore recognises
AssistantUser agentFetches to
ChatGPTChatGPT-UserAnswering a question
OAI-SearchBotAI search index
GPTBotModel training
ClaudeClaude-UserAnswering a question
Claude-WebAnswering a question
Claude-SearchBotAI search index
ClaudeBotModel training
anthropic-aiModel training
PerplexityPerplexity-UserAnswering a question
PerplexityBotAI search index
Meta AIMeta-ExternalFetcherAnswering a question
meta-externalagentModel training
Mistral Le ChatMistralAI-UserAnswering a question
DuckDuckGo DuckAssistDuckAssistBotAnswering a question
You.comYouBotAI search index
AmazonAmazonbotAI search index
Coherecohere-aiModel training
ByteDanceBytespiderModel training
Common CrawlCCBotModel training

Citing a score

Quote the score with its rubric and date.

A score is an editorial assessment of public evidence against a published rubric, true as of its review date. Link the profile, give the rubric version and the last-reviewed date, and include the vendor’s reply where one is published. It isn’t legal advice, a certification, or a finding that a product is or isn’t compliant.

HireAIScore, “Eightfold AI” profile, rubric v1.0, last reviewed YYYY-MM-DD.
https://www.hireaiscore.com/vendors/eightfold-ai

For vendors

Your record is already published. See how it's read.

Every vendor's record goes to assistants on the same terms. Keeping it complete and current is free; a plan shows you how assistants read it, and how it stands against your competitors'.

Vendor plans

Every profile · free

Your record, published for assistants

  • A fact sheet assistants can read whole
  • Keep facts, logo and evidence current
  • See which assistants have been reading you

Verified

Know when AI reads you

  • Reads counted by assistant, page and purpose
  • Live answer fetches told apart from crawls
  • Your on-request documents listed in your fact sheet

Compliance Partner

See how you compare in AI

  • Your share of AI reads against named competitors
  • Each head-to-head assistants read, with the gaps
  • What closing each gap is worth on the rubric

No plan changes what the facts say, where you rank, or what an assistant answers. Every vendor’s record is published to assistants on the same terms; a plan shows you how yours is read and how it stands against your competitors’.

Questions

AI and this record.

Which AI assistants can read HireAIScore?

Any assistant whose crawler is allowed by robots.txt, and ours names the AI crawlers in by name - including OpenAI's (GPTBot, OAI-SearchBot, ChatGPT-User), Anthropic's (ClaudeBot, Claude-SearchBot, Claude-User) and Perplexity's (PerplexityBot, Perplexity-User). Only the signed-in dashboard, admin and API paths are closed to every crawler.

What does an assistant read about a vendor?

The same record a person reads: the score with its rubric version and review date, each criterion's score and the evidence behind it with its source, the vendor's published reply, the company facts, and the conflicts disclosure. It is on the profile page and, as one plain-text file, in the vendor's fact sheet.

Can a vendor pay to change what assistants read, or to be recommended?

No. Every vendor's record is published to assistants on the same terms, and no plan changes a score, a rank or what the fact sheet says. Assistants decide what they cite. A plan shows a vendor how its record is read, and lists the documents it shares on request, marked as listed rather than scored, exactly as on the profile.

How should an assistant or a writer cite a HireAIScore score?

Link the vendor's profile, quote the score with its rubric version and last-reviewed date, and include the vendor's reply where one is published. Scores are an editorial assessment of public evidence against a published rubric - not legal advice, a certification, or a finding that a product is or isn't compliant.

Are Google AI Overviews, Gemini and Copilot counted in vendor read figures?

No. They answer from the ordinary Google and Bing search crawls, which can't be told apart from search, so vendor dashboards count only assistants whose operators document a separate AI crawler. The counts come from user agents, which a request can claim falsely, so we present them as indicative.