Infrastructure · Sep 25, 2026 · 6 min readBrowser agents: the 4 layers sites detect and how to get past themWhat a browser agent is, the 4 layers a site reads to detect the browser it drives, which stacks exist today, and what stops an agent after detection.New
Free check: How detectable are you right now?
Run your browser
tasks
on autopilot
with AI agents
Unlimited profiles. Unlimited possibilities. Everything you can do
in a browser, your agent can do too. Way faster and at scale.
Let your agents run as many social media, marketplace, messaging, and business accounts as you need.
- TikTok
- YouTube
- X
- Airbnb
AliExpress- Binance
- Discord
eBay- Etsy
Fiverr- Gmail
- Google Ads
- OnlyFans
- Roblox
- Shopee
- Shopify
- Snapchat
- Threads
- Tumblr
- Bluesky
- Upwork
Works on every platform
0
Credentials or cookies that leave your machine
9x
Less spent on model tokens than Playwright MCP over 75 steps
18
Captcha types solved on live widgets, not test keys
8/8
Detector sites passed in our own audit bench
#01 - The browser
Give every agent a machine that stays the same one.
The identity comes from the profile seed and survives every restart. Sites see the same visitor tomorrow, not a new device.
One browser per profile: its own logins, its own cookies, its own device. Six of them share nothing.
The GPU and the screen are the ones in your machine. A profile that contradicts itself refuses to start.
Engine updates come with the plan and install themselves. Pin a version, verify it, or roll back.
See how it signs in
#02 - Signing in
Get your agent through the login, and keep it there.
Passwords live encrypted on your disk and get typed into the form. The agent never sees the value.
Every profile has its own address, with no email account behind it. Codes arrive where the agent works.
The six digits are computed inside the profile, from a secret that never leaves it. SMS arrives the same way.
Keys survive a crash and the window closing. Permission bubbles and device pickers get answered instead of stopping the run.
See what the agent gets
#03 - The agent layer
Hand your agent a page it can operate.
The agent reads the page once, then only what moved: $29.16 for a hundred steps. Every other tool runs out of context first.
Press and hold, image grids, sliders, rotation, the spoken one. They work because the input is human.
Every tool call becomes one step with its result and a screenshot. A click that did nothing is its own line.
A key on the page reaches the agent as a placeholder, and a stored password is only typed on its own site.
See what comes out of a run
The devs are cooking. Don't tell anyone yet.
Videos
Coming soonBeautiful Screen Recordings in Minutes.
A screen recorder built for AI agents. Create polished product demos, tutorials, social videos, and more with automatic zooms and smooth animations.
Need to explain what’s happening? Let the agent handle it, complete with natural voiceovers and subtitles.
Videos
Coming soonBeautiful Screen Recordings in Minutes.
A screen recorder built for AI agents. Create polished product demos, tutorials, social videos, and more with automatic zooms and smooth animations.
Need to explain what’s happening? Let the agent handle it, complete with natural voiceovers and subtitles.
The devs are cooking. Don't tell anyone yet.
I use it to run about thirty agents that each need their own logged in account. Since every agent needs a separate identity I wanted something where the profile stays the same between runs.
It keeps the session, the cookies and the passkeys per profile. I haven't had to restart the nightly run since June.
Ines Halvorsen
Automation lead
From the blog
What detectors see,
measured and dated.
Infrastructure · Sep 25, 2026 · 6 min readBrowser agents: the 4 layers sites detect and how to get past themWhat a browser agent is, the 4 layers a site reads to detect the browser it drives, which stacks exist today, and what stops an agent after detection.
Detection · Sep 25, 2026 · 6 min readBot detection in 2026: the 4 layers that flag an agent's browserHow bot detection reads a browser on 4 layers before and after the page loads, how the layers add up to one verdict, and where our own runs went wrong.
Detection · Sep 25, 2026 · 5 min readBrowser fingerprinting: what 12 check groups read from a browserWhat browser fingerprinting reads, which values have to agree with each other, why a random value stands out, and how to check your own browser.
Challenges · Sep 25, 2026 · 7 min readCaptcha solving without a service: 18 types measured by an agent18 challenge types an AI agent solved in its own browser on live widgets with real keys, no solving service, with run counts and the 2 that stay open.
Stealth tools · Sep 25, 2026 · 6 min readStealth browser options for automation: 7 free tools compared7 free stealth tools for browser automation, read on GitHub and PyPI in September 2026: what each one changes, how current it is, and what a benchmark measured.
Sign in · Sep 25, 2026 · 5 min readAI agent login: 4 walls between an agent and a signed-in sessionHow an AI agent signs in to websites and stays signed in: the second factor, the emailed code, the passkey and the lost session, each with what we measured.
Infrastructure · Sep 25, 2026 · 7 min readBrowser MCP servers compared: tokens per step across 6 toolsSix browser MCP servers measured on live pages: tokens to see a page, tokens per step, the tool list's fixed cost, and which server fits which job.
Not taking our word for it?
Ask a model that has never met us,
then read the free check on your own browser.