Get AI access
get_ai_accessUse this tool when the user asks whether ChatGPT, Claude, Perplexity, Gemini or another AI reads their landing page (ランディングページ, ホームページ, Webサイト), which AI crawlers fetched it, or how many visitors AI assistants sent (「AI に読まれている?」「ChatGPT から来た人は?」).
Returns one page's AI access for 7d, 30d (default), or 90d; 180d, 365d, and all need Pro or MAX, as the other long periods do (AI reads were first recorded on 2026-10-25, so their days start there at the earliest; referrals cover the whole period, read from the long-term daily summaries for 180d, 365d, and all).
totals: ai (answer + crawl), answer (an AI opened the page because a person asked it something right then: ChatGPT-User, Claude-User, Perplexity-User…), crawl (AI crawlers: GPTBot, ClaudeBot, OAI-SearchBot, PerplexityBot…), and search (bingbot and Applebot, search crawlers that also feed Copilot and Apple's features, never counted in ai).
verification: the fetches by IP address check, each with the same four numbers. verified: from an IP address the agent's vendor publishes. noList: not checked (the vendor publishes no list, its list has no ranges of the IP address's family or wasn't refreshed for a week, or the fetch was recorded before the check); not checked doesn't mean fake. unverified: the vendor's current list covers that IP address family and the IP address isn't on it. unverified.ai ("self-named only", typically a scanner) is left out of totals and every list; unverified.search (bingbot, Applebot) stays counted, informational only. totals = verified + noList + unverified.search.
byVendor, byToken (each agent's User-Agent token: fetches counted, verified among them, unverified flagged, outside fetches except for search crawlers), byPath (site and path, no query), byDay.
referrals: visitors and pageviews whose utm_source or referrer names an AI assistant (ChatGPT, Perplexity, Claude, Gemini, Copilot, Grok, DeepSeek, Meta AI), with byAssistant. Assistants' apps send no referrer, so this is a floor. startDate and endDate are the days counted; coversPeriod false means a longer period could only be read for its last 90 days. visitorsMethod: unique_in_period (7d, 30d, 90d) or sum_of_daily_uniques (180d, 365d, all: each day's visitors added up, so someone who came on 3 days counts 3 times, as in the other long-term visitor counts). available false means page views couldn't be read right now, not zero.
liveDaysAvailable: today's and yesterday's AI reads are read live; false means they couldn't be read right now, not zero, so the totals leave those two days out. Agents are named by what the requests' User-Agents claimed; where the vendor publishes its IP addresses (OpenAI, Perplexity, Anthropic, Microsoft, Apple, DuckDuckGo, Google), each fetch is checked against them and the IP address is never stored. Only fetches of the live page are counted (not robots.txt or the sitemap). AI reads and referrals are on the workspace's days (fetchDayTimeZone, referralTimeZone; UTC when none is set); referrals for 180d, 365d, and all are on the days the daily summaries are cut in, which follow a zone change once it has finished rebuilding, as get_analytics_summary's long periods do. Right after the workspace's zone changes, AI reads stay in the earlier zone, as fetchDayTimeZone says, until an hourly run re-reads the page; fetchDayTimeZoneSince, when set, is the first day in fetchDayTimeZone, and earlier days are in the zone before. Neither is a page view or a visitor in get_analytics_summary.
Input Schema
| Name | Required | Description | Default |
|---|---|---|---|
| period | No | Time period. Defaults to 30d. 180d/365d/all require Pro or MAX plan. | 30d |
| siteId | Yes | The ID of the page to read AI access for |