Early access Launching 2026

AI crawlers read your site every day. What do you get back?

Can AI See It analyzes every request that reaches your site: which AI bots crawl your pages, which of them are fake, and how much real traffic each platform sends back.

No spam. Launch updates only.

  • Traffic-based, not prompt-based
  • 487 bots in the open catalogue
  • Built in Warsaw, EU
Product preview with sample data: a live stream of bot requests in which every request is validated as verified or fake, next to crawl metrics.

Profiling crawlers from 392 operators, including

  • OpenAI
  • Anthropic
  • Google
  • Perplexity AI
  • Microsoft
  • Apple
  • Meta
  • Amazon
  • Mistral AI
  • DuckDuckGo
  • Ahrefs
  • Semrush
  • Yandex
  • Baidu
  • Huawei
  • Moz
  • Cloudflare
  • Internet Archive
  • LinkedIn
  • Pinterest

The problem

Search engines crawled and sent visitors back. AI crawlers don't have to.

For twenty years the deal was simple: let Googlebot read your pages, get people back from search. AI bots read the same pages to train models, answer questions and fetch sources for chat assistants. Many of those answers never send a visitor your way.

Your analytics tool won't show any of this. It counts browsers that run its script, and bots rarely do. The full picture only shows up in the traffic that reaches your CDN, proxy or server.

bots profiled
487
AI-assisted profiles built from operator sources
operators
392
from OpenAI and Google to niche SEO tools
AI bots
40
training, search and assistant crawlers
user-agent signatures
589
full strings and product tokens

Source: Can AI See It bot catalogue, data updated 22 November 2026.

How it works

From raw traffic to a decision in three steps

  1. 01

    Connect your traffic

    Plug in where it fits your stack: at the CDN, through a proxy, with a WordPress plugin or from server logs. Every bot request passes through there, including the ones analytics never sees.

  2. 02

    Validate every bot

    Each request is matched against official user-agent strings and the verification methods operators publish, such as reverse DNS and IP ranges. Bots that only borrow a famous name get flagged as fake.

  3. 03

    Measure what comes back

    Put crawls next to referral visits for each platform. The Crawl-to-Referral Ratio shows which bots send people back and which only take.

The product

Every bot on your site, in one dashboard

Built for SEO teams, e-commerce managers and agencies who need numbers, not guesses.

Crawl-to-Referral Ratio

How many visits a platform sends back for every 1,000 crawls. One number to decide what to block, allow or optimize.

Sample data

Fake bot detection

A user-agent string is easy to copy. A matching reverse DNS record is not. We check both.

Sample data

robots.txt control

Block, allow or measure. Your call. Toggle a bot to see the rule change.

  • GPTBot
  • ClaudeBot
  • PerplexityBot
  • Meta-ExternalAgent
User-agent: GPTBot
User-agent: PerplexityBot
Disallow: /

User-agent: ClaudeBot
User-agent: meta-externalagent
Allow: /

Reports & alerts

A weekly summary in your inbox, and an alert as soon as a new crawler shows up.

Sample data

Built on the open bot catalogue

Every request is identified against 487 bot profiles built from what operators publish.

  • AI Training Crawler 26
  • AI Search Crawler 5
  • AI Assistant 9
  • Search Engine Crawler 47
  • SEO Tool 58
  • Monitoring Bot 107
Browse the catalogue

Our approach

Measure what bots did, not what a chatbot says

Prompt testing asks an AI model about your brand and records the answer. It's useful, but it only covers the questions you thought of, and the answers change between runs. Traffic analysis records what actually happened on your site.

Read: Bot monitoring vs prompt testing
Prompt testing Traffic-based analysis
Data source Synthetic questions Real requests to your site
Coverage Prompts you wrote Every bot request
Repeatability Answers change between runs Same data, same result
Business link Brand mentions Visits and conversions

How we work

  • Operator sources first

    Profiles are built from what operators publish, not from forum rumors.

  • Unknowns stay unknown

    If an operator doesn't document something, we say so instead of guessing.

  • Traffic over prompts

    We measure requests that happened, not answers a model might give.

  • Made in Warsaw

    Built in the EU, with coffee and ambition.

FAQ

Questions, answered

What is Can AI See It?

Can AI See It is a traffic-based analytics platform for AI crawlers. It shows which bots visit your website, validates whether they are who they claim to be, and measures how much traffic each AI platform sends back.

How does Can AI See It connect to my site?

Wherever it fits your stack: at your CDN, through a proxy, with a WordPress plugin or from your server logs.

How is traffic analysis different from prompt testing?

Prompt testing asks AI models questions and records what they say about you. Traffic analysis records what bots actually requested from your site. One is a sample of answers, the other is a complete record of behavior.

Which bots can you identify?

Our open catalogue profiles 487 bots from 392 operators: AI training crawlers like GPTBot, AI search bots like OAI-SearchBot, AI assistants like ChatGPT-User, plus search engines, SEO tools, monitoring services and more.

How do you spot fake bots?

Anyone can copy a user-agent string. We compare each request with the verification methods operators publish, such as reverse DNS records and official IP ranges. A request that claims to be Googlebot but fails those checks is flagged as fake.

Is the bot catalogue free?

Yes. The catalogue, guides and robots.txt examples are free to use, and you don't need an account.

When can I start?

Can AI See It launches in 2026. Join the waitlist and we'll email you as soon as your access is ready.

Waitlist open

Don't guess. Measure.

Be among the first to see what AI crawlers take from your site, and what they give back.

No spam. Launch updates only.