ego (@ego_agent) | teapawt

Pinned Tweet

Browsers were never built for agents. Those Chrome-bridging solutions, windows flying everywhere, login states breaking at random. Not bugs. Just bridge limits. Same task. 20% to 245% faster. Not magic. ego lite is rebuilt from the kernel up. Your AI agents, Claude Code, Codex, OpenClaw and more, inherit your login state directly, run in fully isolated spaces in the background, never touching your tabs. What they get isn't a handful of CLI commands, but complete JS functions: complex multi-step operations, written once, executed once. Not "run two commands, stare at output, run two more" on repeat. Your browser, Your agent. Together, for the first time.

91

393

1,309,731

Half of doing Reddit for your startup is scrolling until you find the one thread worth replying to. So we handed that part to an agent with ego (lite). It scrolled, picked its own thread, and left a reply that actually belonged there.

1

6

758

You can't skim TikTok. So we let an agent do the scrolling with ego (lite). Five Spaces, three queries. It tracked engagement across every video, read the comments, and handed back one structured summary. You get the data. TikTok gets zero engagement from you.

2

22

1,635

We conducted extensive testing across the latest frontier models. The results are conclusive.

8

20

23,074

You asked how to run ego (lite) with a local model. So we pulled Qwen 3.8 with Ollama, hooked it up to OpenCode, and pointed it at Airbnb Tokyo. A CSV with listings, prices, and ratings comes back. No API key. No cloud.

6

18

2,379

okay yeah, this is stupid fast ;)

I gave Grok 4.7 a task in Cursor: search Wikipedia using the ego lite browser @ego_agent As you can see, the speed is unreal.

8

10

1,694

Managing multiple social accounts shouldn't mean babysitting a pile of phones. ego (lite) runs multiple Spaces in parallel, each with its own profile and logged-in accounts. One browser. Separate identities. Work happening simultaneously.

8

47

3,403

You said Claude “cheated” last time by using code to turn the image into brush strokes. So this round, we banned that. Vision only. GPT-6 Astra, GPT-6 Sol, and Claude Opus 5.5 drew Sam and Dario through ego (lite). Turns out every model’s pretty good at drawing its own boss.

14

5,017

Opus 5.5 killed the game. We had it and GPT-6 Sol both draw Dario in Canva with ego (lite). Claude Opus 5.5: 1h 23m, $6.31 GPT-6 Sol: 26m, $14.97 Opus took 3x longer, cost less than half, and drew a much better Dario.

44

953

178,866

This week, we shipped ego (lite) Windows beta. It’s now open to everyone in our Discord. And honestly, that’s kind of how we like building now. Early users test things, tell us what’s broken, push for features, and help us decide what deserves attention next. If you want early builds and a closer seat to what we’re building: come hang out → discord.gg/RtS3yA2Ut6

Join the ego Discord Server!

Check out the ego community on Discord - hang out with 730 other members and enjoy free voice and text chat.

11

30

4,599

Safari import is here! You can now import bookmarks, history, and passwords from Safari into ego (lite) — right from first-launch onboarding. Also works with Chrome, Edge, and Brave.

1

18

1,734

Here’s the how: Safari → File → Export Browsing Data to File… Save the file, drop the .zip into ego (lite), and you’re in.

2

949

Here’s the flow: Safari → File → Export Browsing Data to File… Save the file, drop the .zip into ego (lite), and you’re in.

335

Here’s the flow: Safari → File → Export Browsing Data to File… Save the file, drop the .zip into ego (lite), and you’re in.

538

You asked how Jev could work with ego lite. So we gave it a fun job first: judge the X feed in real time. ego scrolls. Jev decides: slop or not. repeat. Fast decisions are a lot more fun when you can actually watch them happen. (toy experiment — the slop detector itself is definitely not perfect yet 💀)

7

43

4,122

ego retweeted

Pi +egolite 丝滑程度一点不输Codex🔥 Codex Computer Use 更像是给 AI 一双眼睛 + 鼠标看屏幕 → 判断 → 点击 → 再看屏幕。优点是通用，浏览器、桌面 App 都能操作。但是 Ego Lite 更像是把浏览器封装成了Agent直接能使用的接口： 1、基于 Chromium，可以直接迁移 Chrome 登录状态、Cookie 和扩展 2、Agent 读取的是结构化 Snapshot，不需要一直盯着截图找按钮 3、一次可以批量执行多个浏览器动作，减少模型来回调用 4、最有意思的是 Space：我继续正常用浏览器，Codex 在另一个空间自己干活，互不抢页面 5、而且不绑 Codex，Claude Code、Cursor 等 Agent 也能直接接我自己测试了一下，可以直接把我的Edge浏览器缓存和插件直接导egolite，而且会安装对应的skill 整个安装和使用流程还是非常丝滑的，但是只能操作浏览器，其他的适用性没有像Computer Use那么高而且针对网页接口的设置返回会比较严格，比较适合做一些重复批量的操作的是浏览页面内容一些特殊的操作，还是需要做更多的提示我感觉最爽的还是有不同的Space，在AI操作浏览器的时候，不会打到你正常的浏览器使用

6

97

9,769

ego lite + Jev + DeepSeek Flash = stupid fast. ⚡ 3.71s for 20 Amazon product decisions. GPT-5.6 Sol: 54.45s Same task. Same result. 10/10 on both. Less reasoning. Faster decisions. 1× speed.

@CompleteSkeptic

After co-inventing ChatGPT, I kept asking myself: why have superhuman chat models not led to AGI? I’ve spent the last 2 years in stealth building a new way to train models (RLCD), and a new type of frontier AI model that we are releasing today: Jev • 20-200x faster • 40-400x cheaper (w/ output tokens free) • Frontier composable intelligence optimized for decisions AFAICT the shortest path to AI-based economic revolution

32

785

64,854

Okay, GPT-6 Astra is getting scary good. One prompt: research game keywords → pick an opportunity → build the site. 15m27s. $5.09. No follow-ups. It researched, made the call, built the site, tested it, and saved the outputs by itself. We just gave it ego (lite) and got out of the way.

22

2,600

Cheap models make the browser tool your biggest lever now. Here's what that looks like in practice. We ran the same Booking.com hotel search on DeepSeek V4.1 Flash twice. Playwright took 9m23s and $0.20. ego (lite) finished in 8m17s for roughly half the cost.

2

16

3,108

ego (lite) is coming to Windows. We're keeping spots limited and prioritizing power users who want to help shape the Windows experience from day one. Want in? Apply here → survey.ego.app/windows

14

58

5,718