Metrics
How often AI assistants read Zeitwire for the people asking them, which AI and search companies visit, clicks from AI answers, and scrapers turned away.
Counting since October 27, 2026 at 8:00 PM UTC.
16 agents in the past hour2 AI assistant reads in 24h
Last 7 days, refreshed every minute
Which agents
Crawler visits in the last 7 days. Verified visits were confirmed by the operator's published IP ranges, reverse DNS, or Cloudflare. Self-identified visits only sent the agent's user agent.
| Agent | Operator | Verified | Self-identified |
|---|---|---|---|
| OAI-SearchBot | OpenAI | 19 | 0 |
| Googlebot | 17 | 0 | |
| Bingbot | Microsoft | 8 | 0 |
| ChatGPT-User | OpenAI | 2 | 12 |
| DuckDuckBot | DuckDuckGo | 1 | 0 |
| Perplexity-User | Perplexity | 0 | 1 |
What agents read
The 10 most-read pages for AI assistants and search crawlers in the last 7 days. A page is named only when a verified crawler read it; other reads count toward their section.
| Page | Reads |
|---|---|
| Homepage | 11 |
| Other pages (other pages) | 6 |
| About pages (other pages) | 3 |
| Pulse Check (other pages) | 3 |
| codex | 3 |
| grok-bot | 3 |
| @karpathy | 2 |
| Articles (other pages) | 1 |
| @aakashgupta | 1 |
| @askalphaxiv | 1 |
AI referral clicks
Readers who arrived from an AI answer in the last 7 days, by assistant.
| Assistant | Clicks |
|---|---|
| ChatGPT | 0 |
| Perplexity | 0 |
| Claude | 0 |
| Gemini | 0 |
| Microsoft Copilot | 0 |
Scrapers turned away
Requests answered with 403 in the last 7 days, by reason.
| Reason | Requests |
|---|---|
| Scraping library user agents (python-requests, curl, Scrapy, and others) | 1 |
| Probes for exposed files and admin pages (/.env, /wp-admin, and others) | 0 |
How we count
- An assistant read is a page an AI assistant fetched because a person asked it something just then: ChatGPT-User, Claude-User, Perplexity-User, DuckAssistBot, MistralAI-User. Search crawlers build indexes (Googlebot, Bingbot, DuckDuckBot, YandexBot, Applebot, OAI-SearchBot, Claude-SearchBot, PerplexityBot).
- A visit is verified when Cloudflare's verified-bot signal, a reverse DNS lookup confirmed by a forward lookup, or the operator's published IP ranges confirm the crawler. Anyone can send a crawler's user agent, so a visit with only that is counted as self-identified.
- Agents counts separate AI agents seen in the past 60 minutes: AI assistants and AI search agents, verified or self-identified. Google, Bing, DuckDuckGo, and Apple search crawlers are not counted. One session is one agent from one network address and user agent.
- AI training crawlers are not counted anywhere on this page.
- An AI referral click is a page load that arrives from chatgpt.com, perplexity.ai, claude.ai, gemini.google.com, copilot.microsoft.com, by Referer or utm_source.
- Scrapers turned away are requests the site refuses: HTTP libraries on reader pages, and probes for files such as /.env.
- Hourly counts store the agent, the section of the site, and the page for verified assistant and search crawlers. To count separate AI agents, each AI assistant or AI search visit gets a session ID: a keyed HMAC-SHA256 hash of the IP address, user agent, and agent name, using a random salt that changes every UTC day. The salt is kept apart from the hashes and deleted when the day ends, and session IDs are deleted after 7 days. IP addresses, user agents, and query strings are never stored.
- The last 7 days means the 168 hours before the latest refresh, and the change compares them with the 168 hours before that. Every figure on this page and in the header pill comes from one reading that refreshes every minute.
- Building an agent? See /agents for the APIs and discovery files.