UPDATES
Keller Maloney
Unusual - Founder
Mar 26, 2026
AI agents are increasingly visiting websites instead of humans to collect information, answer questions, and even take actions for humans.
This is an “AI impression”: where an AI model forms or updates an opinion about a brand based on its visit to its site. These impressions shape what the agent knows, and therefore how it talks about, the brand
The problem is that no one can see them. Traditional analytics tools like Google Analytics, PostHog, and Amplitude don't detect AI agent traffic because most agents don't execute JavaScript — they fetch content directly or use headless browsers. So marketing teams investing in positioning, sharpening their narrative, and publishing better proof have no way to tell if AI models are actually picking up on it.
So we built a free analytics tool that can detect AI agents!
Introducing Agent Analytics: a free, open-source web analytics tool that shows you when AI agents visit your website, which pages they read, and how traffic changes over time.
What It Does
Agent Analytics detects when AI models search the web during a conversation and land on your site. It tracks 14 different AI crawlers across 8 organizations:
OpenAI: GPTBot, ChatGPT-User, OAI-SearchBot
Anthropic: ClaudeBot, Claude-SearchBot, Claude-User
Google: Google-Extended, Google-CloudVertexBot
Perplexity: PerplexityBot, Perplexity-User
DeepSeek: Deepseekbot
Meta: FacebookBot, meta-externalagent
Mistral: MistralBot
The dashboard shows you:
Total agent requests and how they trend over time
Which agents are visiting (OpenAI vs. Anthropic vs. Perplexity, etc.)
Which pages get crawled most frequently — so you can see exactly what content AI models are pulling from when they answer questions about your category
Data transferred — how much of your content agents are actually consuming
Why This Matters
We've seen companies like Vercel and Mintlify report that a significant chunk of their web traffic now comes from AI agents rather than humans. TailwindCSS recently had to restructure because their monetization model relied on human website visits, but now the majority of their docs traffic is coding agents.
Even if you're not selling a dev tool, AI agents are almost certainly visiting your site. And the pages they visit are a direct signal of the conversations buyers are having with AI about your space — which pages get pulled into context when someone asks ChatGPT about your category, and how that changes as you invest in content and positioning.
Before Agent Analytics, there was no way to see this. You could publish a case study, update your positioning page, or write a comparison article — and have zero visibility into whether AI models actually started pulling from it. Now you can.
How It Works
Agent Analytics connects to Cloudflare's analytics to identify and track AI bot requests to your domain. Cloudflare sits between your visitors and your server, which means it sees all traffic — including the requests from AI agents that never trigger JavaScript-based analytics.
Setup takes about five minutes:
Sign up at analytics.unusual.ai
Connect your Cloudflare account with an API token
Agent Analytics pulls your historical data — you'll see the last 30 days of agent traffic immediately
Your Cloudflare API token is encrypted and only decrypted server-side. We never expose credentials to the browser.
A background sync runs every six hours, pulling the latest agent traffic data from Cloudflare's GraphQL API into your dashboard. No code changes to your site are required — if your site is behind Cloudflare (and most are), it just works.
It's Free and Open Source
Agent Analytics is completely free. No credit card, no trial period.
The code is open source under the MIT license. You can inspect exactly what it does, self-host it if you prefer, or contribute to it.
We built this because we think every company should be able to see how AI models interact with their content. It's a foundational piece of visibility that the existing analytics stack doesn't provide.
How This Fits Into the Bigger Picture
Agent Analytics tells you which pages AI models are reading. It doesn't tell you what they conclude after reading them — whether they recommend you, how they position you against competitors, or what misperceptions they're carrying into buyer conversations.
That's the problem Unusual solves. Agent Analytics is the traffic layer — it shows you AI is reading your content. Unusual is the perception layer — it shows you what AI believes about your brand and what to do about it.
You don't need Unusual to use Agent Analytics. But if you're watching AI agents hit your site and wondering what they're actually telling buyers about you, that's where the two connect.
Start tracking for free: analytics.unusual.ai
Questions? Reach out at support@unusual.ai.
