ego (lite) launches Windows beta version, supporting independent browser workspaces for multiple Agents
BIBIBI
AT A GLANCE
The report states that AI Agent browser ego (lite) has launched Windows beta version, enabling various Agents to operate web pages and execute tasks in parallel.
Article
ego lite launches Windows: Claude and Codex share a browser, with multiple Agents working in parallel in the background
Beating AI flash report:AI Agent dedicated browser ego (lite) officially launches Windows beta version, having previously supported only macOS.
Users can use `ego-browser`Skill to allow existing Agents such as Claude Code, Codex, and Cursor to operate the browser directly.
ego (lite) is built on Chromium and supports one-click import of Chrome bookmarks, extensions,Cookie and login states. Agents can directly operate already logged-in sites such as Gmail, X, and enterprise back-office systems to complete information gathering, form filling, and web testing.
Each Agent has an independent Space workspace and can execute multiple tasks simultaneously without taking over tabs currently being used by the user. For steps such as CAPTCHAs and payments, users can take over at any time.
It also changes how Agents read and operate web pages. The browser directly generates a streamlined page structure, marking elements such as buttons and input fields, thereby avoiding sending the entire page HTML to the model. Agents can write multiple actions into a JavaScript sequence for continuous execution, reducing tool calls and Token consumption. In internal testing across 4 browser tasks, the official claim is that it was up to 3.45 times faster than Vercel's agent-browser.
Original link https://m.theblockbeats.info/flash/372678
Key points
01
Beating AI flash report states that ego (lite) officially launched Windows beta version, having previously supported only macOS.
02
The original article states that users can use ego-browser Skill to allow Agents such as Claude Code, Codex, and Cursor to operate browsers.
03
ego (lite) is built on Chromium and supports one-click import of Chrome bookmarks, extensions,Cookie and login states.
04
The original article states that each Agent has an independent Space workspace and can execute multiple tasks simultaneously, while users can take over for steps such as CAPTCHAs and payments.
05
The original article states that the browser generates a streamlined page structure and marks elements such as buttons and input fields, while Agents can also continuously execute multiple actions using a JavaScript sequence.
06
The original article states that, in internal testing across 4 browser tasks, it was up to 3.45 times faster than Vercel's agent-browser.
AI-assisted interpretation
The following is analysis, separate from reported facts. Verify important claims independently.
This is a browser beta version that enables AI Agent to operate web pages directly. It emphasizes reusing existing login states, assigning separate workspaces to different Agents, and reducing tool calls and Token consumption through streamlined page structures and batch action execution.
Why it matters to readers
If these features are actually available, Agents such as Claude Code, Codex, and Cursor could handle tasks including information gathering, form filling, and web testing without occupying the user's current tabs.
Beginners can understand it as a browser for AI assistants:AI can click and fill in forms on users' behalf, but users can still take over sensitive steps such as CAPTCHAs and payments.
Risks and unknowns
Windows The scope of availability, installation method, and actual stability of the beta version are not specified in the evidence.
Actual compatibility with Claude Code, Codex, and Cursor, as well as permission controls, has not been independently verified.
The security protections involved in importing Cookie and login states are not specified in the evidence.
3.45 The performance figure comes from official internal testing; the evidence does not provide testing methodology or independent verification materials.
Related Developments
Loading event timeline…
Related concepts
AI Agent
that can understand information, reason, and autonomously call tools and perform actions after obtaining the requisite permissions AI system.
