Aether (@aetherruntime) header
Aether (@aetherruntime) profile picture

Aether turns your AI subscription into a dev team. Never waste time multitasking agents again.

San Francisco
Joined July 2026
Photo by Aether (@aetherruntime)Photo by Aether (@aetherruntime)Photo by Aether (@aetherruntime)Photo by Aether (@aetherruntime)Photo by Aether (@aetherruntime)
Replying to @aetherruntime
With Aether I can keep building while 10 shots deep in the club, that’s actually when the real ideas form
1
1
28
Replying to @XanderBartlettt
I prefer a vodka redbull but same idea💪🏽
1
1
13
Aether retweeted
Replying to @neerajjj6785
1
1
21
Replying to @thsottiaux
I WANT PROOF! When a task finishes, show me the app running in a clean env: test output, preview, screenshots. "Done" should come with receipts
2
565
Replying to @moiiikaaa
Leaderboards won't answer this. Give all three the same real task from your repo, in the same clean environment, and compare did it pass tests, how many retries, what it cost, etc. the winner changes per codebase more than people think
1
56
Replying to @mattpocockuk
Run the same retro on outcomes too: "find every session where the agent said done and I still had to fix it." that list is basically a map of your missing tests.
162
Replying to @rauchg
No actually AI does not make trust free. Generating code is heading to zero but knowing it actually works isn't. Verification is the last expensive thing in the loop.
3
230
We don’t talk enough about code review and it’s probably why everyone’s doing it wrong. Having another agent review your PR is a meaningless rubber stamp. Even if you use a different model, a reviewer agent reads code, but it doesn't run it. Two agents agreeing is not evidence of anything. If you truly want it verified the only way is an isolated devbox with real tests, previews, and logs. If you’re a serious developer get serious about review.
1
3
709
this is why Aether was built Aether has the best computer use tools of any platform. The agent gets tools to test its changes in a browser, terminal, desktop app, or iOS or Android mobile device
A: an agent with access to a debuggable/clickable dev server B: one without I'd argue that the difference between A and B is bigger than most models Fix the environment your agent runs in
299
Every time I see someone talking about how much they love Devin I eventually find out they’re somehow biased. either they’re an employee, investor, paid affiliated, or in this case, possibly all three yet I have never once met someone in real life who uses it. curious.
No product has changed the velocity of my team’s ability to ship more than @cognition They have a new program for startups to get up to $65K in credits. Use my code and apply here: devin.ai/startups?ref=dfs-ma…
58
12
2
458
127,290
A reviewer that reads your diff gives you an opinion. A reviewer that runs your code gives you a screenshot of the thing that broke. We built the second one.
220