Turns out ego (lite) is everyone’s type.
Claude, GPT, and DeepSeek each ran the same research task across three browsers. Then we asked which one they’d commit to.
3 out of 3 chose ego (lite). 💍💍💍
The browser war is over. The group chat has begun.
Receipts in the thread 👇
New intern started today.
Researched 5 AI products end to end, tested each one hands-on, wrote the full playbooks in 25 minutes.
Her name is Cindy. She runs on our existing subscription. ego (lite) is why she never had to sign in once.
Good hire, @DashHuang
DeepSeek’s Super-agent × ego (lite) is is a killer combo.
3 GitHub issues in.
3 workers in parallel, each with its own signed-in browser.
3 PRs back in 6 min, each with before/after screenshots.
The whale brings the brain. ego (lite) brings the hands.
DeepSeek is basically free next to Opus 🎃
Both built a jelly jack-o'-lantern in CodePen with ego (lite), then poked it, carved it and fixed it.
DeepSeek V4.1 Flash 1017: 13m 12s, $0.13
Claude Opus 5.5: 14m 29s, $3.51
Same task. About the same time. 27x the bill.
Without logged-in sessions, your agents are cooked.
Most agent browsers start with an empty profile, so your agents hit a login page everywhere.
ego (lite) runs them with the logins already in your browser.
Same 7 tasks: 1 done in Playwright, all 7 in ego (lite).
ego (lite) is probably the most underrated browser agent atm
my usual multi-step QA task that will take browser use multiple trial and errors in python and takes cua driver 3x as long, is executed flawlessly in ego lite
however i also think it's unfair to compare ego lite to cua and browser use since ego has a closed source browser which does the heavy lifting bts
simply comparing apples to oranges
and whatever ego is doing bts with their browser, where all the magic lies, is a blackbox
atm what's on their github is just their sdk and skills