# egolite: set up agent browsing and inspect a parallel test

Follow the recorded Mac setup, watch separate agent spaces and keep account access and benchmark scope explicit.

[HTML](https://lukejbyrne.com/library/guides/dpl4ruhrr9w/)
Content language: en
Author: Luke Byrne
Video published: 2026-06-22
Guide published: 2026-11-02

[Watch the video](https://www.youtube.com/watch?v=dPL4RuhrR9w)
Video language: English

## Understand the browser and its access

The June video introduces egolite, also called ego browser, as a Chromium-based browser for agents. Luke says it replaced the Claude Chrome extension for his own work. The original destination is [ego browser](https://link.ego.app/ai_luke).

The documented setup downloads the Mac app, moves it to Applications and imports Chrome profiles, with no new account required in the recording. Imported profiles let the agent use existing website sessions without an API key. That is access to real logged-in accounts, not a separate demonstration identity. Decide which account context belongs in a task before giving the agent control.

1. Follow the setup section at 1:38 and the login explanation at 2:01.
2. Inspect which imported account is active before assigning a website task.
3. Keep the task specific about what information to read and what changes, if any, are intended.

## Watch the parallel spaces

The source describes AI spaces as separate tab groups in which multiple agents can work at once. Luke shows following the work, taking control and handing it back. Separate spaces make activity easier to observe, but do not themselves prove every agent has selected the correct account or finished its task.

1. Use the spaces section at 2:23 and control handoff at 3:12 to understand the demonstrated interaction.
2. Follow the YouTube data example at 3:59 and inspect the information actually returned.
3. For a parallel test, give each task a clear expected result and review the evidence separately.

## Keep timing and correctness together

The video reports a parallel end-to-end SaaS test and a comparison with the Claude Chrome extension. Its description claims 20% faster overall and 3x faster on the actual work. Those are reported results from this recording, not a guaranteed improvement for another website, machine or agent.

The description also discusses layout changes and working with coding agents that can install a skill, including Claude Code, Codex and Cursor. It does not provide a reproducible universal reliability benchmark. Review completed actions, wrong turns and the resulting data alongside elapsed time before deciding whether this route suits your own work.

## The takeaway

Use the recorded setup to understand the workflow, then verify account context, task completion and timing on your own bounded job.

## Workflow resources

- [link.ego.app](https://link.ego.app/ai_luke)

## Chapters

- [0:00 Claude browsing the web in parallel](https://www.youtube.com/watch?v=dPL4RuhrR9w&t=0s)
- [1:18 What egolite actually is](https://www.youtube.com/watch?v=dPL4RuhrR9w&t=78s)
- [1:38 Set it up in under a minute (free)](https://www.youtube.com/watch?v=dPL4RuhrR9w&t=98s)
- [2:01 Let Claude use your own logins](https://www.youtube.com/watch?v=dPL4RuhrR9w&t=121s)
- [2:23 Parallel AI spaces in one browser](https://www.youtube.com/watch?v=dPL4RuhrR9w&t=143s)
- [3:12 Following along & taking control](https://www.youtube.com/watch?v=dPL4RuhrR9w&t=192s)
- [3:59 Demo: YouTube keyword data, no API key](https://www.youtube.com/watch?v=dPL4RuhrR9w&t=239s)
- [4:44 A parallel end-to-end test on my SaaS](https://www.youtube.com/watch?v=dPL4RuhrR9w&t=284s)
- [5:35 What the data actually revealed](https://www.youtube.com/watch?v=dPL4RuhrR9w&t=335s)
- [6:18 egolite vs the Claude Chrome extension](https://www.youtube.com/watch?v=dPL4RuhrR9w&t=378s)
- [7:09 Watching the Chrome extension struggle](https://www.youtube.com/watch?v=dPL4RuhrR9w&t=429s)
- [9:22 The results: 20% faster, 3x on real work](https://www.youtube.com/watch?v=dPL4RuhrR9w&t=562s)
- [10:45 Why this is the future of AI browsing](https://www.youtube.com/watch?v=dPL4RuhrR9w&t=645s)

## Sources and limitations

Written companion prepared 1 November 2026 from this exact published video description and its chapter sequence. The full recording was not independently watched or transcribed for this article, and the products and reported results were not retested. Checklists are editorial adaptations of the documented workflow. Model availability, interfaces, prices and offers may have changed since publication; use the retained original resources for current requirements.

[Original video](https://www.youtube.com/watch?v=dPL4RuhrR9w)