Perplexity’s Comet and OpenAI’s Atlas now account for roughly 71% of all observed agentic browsing activity — traffic where an AI, not a human, is doing the clicking. Most of it isn’t buying anything yet: nearly 70% of that activity is browsing, searching and reading, not completing purchases. But it’s growing fast in specific sectors, and it arrives at your site looking almost exactly like a real visitor. Here’s what’s actually happening and what it means for your SEO and analytics.
Quick answer: AI browsers are standalone browsers (like ChatGPT Atlas and Perplexity’s Comet) where the AI itself navigates the web on a person’s behalf — clicking links, reading pages, filling forms — instead of a human doing it manually. As of early 2026, Comet drives about 48% and Atlas about 21% of observed agentic browsing activity, a combined 71%. Most of this traffic browses and researches rather than checks out, concentrated in media (46%), ecommerce (38%) and travel (14%). Because these agents use normal-looking user-agent strings and cookies, they’re hard to separate from human traffic in standard analytics — which means most sites are already getting AI-browser visits without knowing it.
What an “AI browser” actually is
An AI browser is different from an AI chatbot that occasionally cites a web page. Tools like OpenAI’s Atlas and Perplexity’s Comet are full browsers where an AI agent actively navigates: it opens tabs, clicks through multi-step flows, and reads rendered pages the way a person would — not just a search index. That’s a meaningfully different kind of “visitor” than the crawlers site owners are used to thinking about.
The numbers behind the shift
- 71% combined share: browser-based agents, led by Perplexity’s Comet (48.12%) and OpenAI’s Atlas (21.33%), account for roughly 71% of observed agentic traffic.
- 69.57% is research, not transactions: the large majority of agentic activity touches product listings, articles and search results — browsing and comparing, not completing checkouts.
- Media leads adoption at 45.62% of agentic traffic, followed by ecommerce at 38.20% and travel at 14.12%.
- Growth is uneven by sector: government/federal services saw agentic traffic jump 254% month-over-month and SaaS grew 41.5%, while media (+13.3%) and ecommerce (+6.5%) grew more steadily. Education contracted sharply (-96.8%).
- They’re hard to detect: these agents arrive with user-agent strings, cookies and session patterns that closely resemble human browsing, which complicates identifying and measuring them with standard tools.
Where agentic traffic is concentrated
| Sector | Share of agentic traffic | Month-over-month growth |
|---|---|---|
| Media | 45.62% | +13.3% |
| Ecommerce | 38.20% | +6.5% |
| Travel | 14.12% | — |
| SaaS | — | +41.5% |
| Government / federal services | — | +254% |
| Education | — | -96.8% |
What to actually do about AI browsers right now
- Make sure your pages work without heavy client-side rendering tricks. Agentic browsers read rendered content, but clean, well-structured HTML and clear headings still make it far easier for an agent to extract accurate information from your page.
- Don’t assume your analytics already separate this traffic. Because agentic sessions look like normal browser sessions, most GA4 setups are currently blending AI-browser visits into “direct” or generic traffic — see how to track AI referral traffic in GA4 for the fix.
- Prioritize this if you’re in media, ecommerce or travel. These three sectors already account for the vast majority of agentic activity — if you’re in one of them, this isn’t a future problem.
- Keep product and content data accurate and complete. An agent researching on a user’s behalf will move on quickly if your page doesn’t answer the question directly — the same discipline behind ranking in agentic commerce.
- Watch server load and bot-management rules. Agentic browsers can trigger the same rate limits and bot challenges built for scrapers — worth auditing so you’re not accidentally blocking legitimate AI-assisted visitors.
Why this matters even though most agents aren’t buying yet
The 69.57% research-not-transaction figure is the most important number here. It means agentic browsers are currently doing to individual websites what search engines have always done — driving discovery and comparison — except the “search engine” is now inside the browser itself, reading your page directly rather than just indexing it. That has a direct GEO implication: how an AI browser reads and represents your page during that research phase can shape whether a purchase, booking or sign-up happens later, even if the agent isn’t the one clicking “buy.”
What this means for your business
You almost certainly already have agentic browser traffic hitting your site — it’s just invisible in a standard analytics setup. Given that media, ecommerce and travel already account for the majority of this activity, treating it as a “wait and see” category is a mistake for most of our clients. We help businesses get the two things that actually matter here in place: content and product data that reads clearly to an AI agent, and measurement that separates this traffic so you can see whether it’s converting.
Frequently asked questions
An AI browser is a standalone web browser where an AI agent navigates on a person’s behalf — clicking links, reading pages and completing multi-step tasks — rather than a human manually browsing. ChatGPT Atlas (OpenAI) and Comet (Perplexity) are the two leading examples as of 2026.
Browser-based AI agents, led by Perplexity’s Comet (about 48%) and OpenAI’s Atlas (about 21%), account for roughly 71% of all observed agentic browsing activity as of April 2026 data.
Mostly not yet. About 69.57% of agentic browsing activity is research — browsing product listings, reading articles and running searches — rather than agents autonomously completing checkout.
Usually not by default. Agentic browsers arrive with user-agent strings, cookies and session patterns that closely resemble human browsing, so standard GA4 setups typically blend this traffic into direct or generic sessions unless you specifically configure tracking to separate it.
Media (45.62% of agentic traffic), ecommerce (38.20%) and travel (14.12%) account for the large majority of current agentic browsing activity, making them the highest-priority sectors for adapting content and measurement today.
Get ready for the AI browser era
AI browsers are already reshaping how people find and evaluate businesses online — and most sites aren’t structured for it. Get a free AI-readiness audit and we’ll show you exactly how well ChatGPT Atlas, Perplexity Comet, and other AI agents can find, understand, and recommend your site, then hand you a prioritized plan to fix the gaps. Ready to get ahead of the shift? Talk to our team and we’ll take it from there.
Related Reading
- Zero-Click Searches in 2026: What the Data Says About Where Your Traffic Went
- Schema Markup in 2026: The Structured Data AI Search Engines Actually Use
- Google’s New AI Search Guide: What Actually Changes for SEO in 2026
Riya Bhardwaj
Lead Marketing Strategist
Leading content and growth initiatives with a focus on search visibility, audience engagement, and measurable business outcomes. Specialised in SEO, Generative Engine Optimization (GEO), AI search optimisation, and performance-driven content marketing. Passionate about transforming market insights into scalable content strategies that strengthen brand authority and drive sustainable digital growth.