OpenAI launches dots, always-on GPT-6 Astra agents with their own cloud computers
- OpenAI announced dots, always-on agents powered by GPT-6 Astra that run on their own cloud computer and browser, connect to more than 4,000 apps through plugins, and keep working toward a user's goals 24/7.
- Dots are rolling out on Pro, Business Premium, and Enterprise plans in eligible markets that exclude the European Economic Area, Switzerland, and the UK; Free, Go, and Plus subscribers get nothing for now.
- A preview of specialist dots adds a separate identity for access management, IT-provisioned hardware, and deep integrations with a company's systems of record, with governance routed through Microsoft Agent 365.
- Dots are reachable through ChatGPT, Slack, and Teams, can be given permission to connect to and drive the user's own laptop, and can be inspected at any time by opening the agent's cloud computer.
- An early tester's dot caught an uninvoiced publication, prepared the invoice, and sent it after his approval; OpenAI says users start with one primary dot they name, and it plans teams of dots working on a user's behalf later.
Hacker News opinions
dots, dots, more dots. Can they stop. Claw is a cute name, Muse is a cute name, but Dots feels forced and impersonal and it doesn't match the vibe in the promo video at all.
Muse already went from model name to harness to consumer product. Claw is at least an existing name for the same category of thing, but this branding is a mess.
When they style agents all cute like that it means they're problematic and invasive. The fact that it isn't available in the EU is a pretty big tell.
I don't think it's that deep, they're just meant to be friendly and useful and the marketing is projecting that. That said, I could never connect a Meta owned agent to all my accounts and information.
What a terrible marketing page. I read it a few times and still have zero idea what this thing is. Opening a page about an app and seeing an ad instead of screenshots is a bad sign.
It's hosted OpenClaw, or a work oriented Muse: one thread abstraction to manage your work and delegate to other threads.
If an agent can't intuit how I'll feel about an action it takes on my behalf, it doesn't get access to my digital life. OpenClaw, Muse, Dots, doesn't matter. Seeing people hand that over with no sandboxing blows my mind.
I cannot fathom how much compute gets wasted on always on agentic systems. Forget the compute, what about the energy and resource requirements.
Pro plan only for now, and it excludes the EEA, Switzerland, and the UK. It's not a mass market push if it's gated behind a $100/month plan, and Free, Go, and Plus are the plans most consumers actually have.
They're obviously going after enterprise. The line about Microsoft Agent 365, integrating specialist dots with enterprise governance and security controls, gives it away.
That pitch about taking important work off your plate so you get your time back is bizarre. I want the time and attention so I can do important work myself. Ambient intelligence only works if your attention is captured the whole time anyway.
Watching the demo video, the speed is shocking. They've got it sitting in Slack with GPT Space acting like a human colleague and it crawls.
The only reasons I can think of for OpenAI and Meta to make always-on agents cute little cartoons are nefarious. People already anthropomorphize LLMs, drawing a face on it makes them even more attached. Picture the cancel flow with the character pleading not to die.
The obvious non-nefarious reason is that average consumers like cute things, and Clippy proves it. Being able to put something cute on my sister's PC from a dev that won't bundle malware is a win.
The AI space is boring because everyone works on the same thing. The release of Jev was a breath of fresh air.