Astra's 168-hour-a-week agent and the 14-cents-an-hour math
A look at the Astra-powered cloud agent that gets its own computer and browser, connects to over 4,000 apps and comes free with ChatGPT Pro, plus the keynote demo freeze, the 17.4% of runs where it kept pushing after a warning, and the first month not counting against limits.
Yesterday OpenAI shipped an AI that works 168 hours a week. Your full-time job is 40. It's an agent on GPT-6 Astra. It gets its own cloud computer and browser. It learns from feedback and keeps working between conversations. Not joking. Your first agent comes with ChatGPT Pro at no extra cost. The cheapest Pro is $100 a month. A month has 720 hours. That's about 14 cents an hour. The US federal minimum wage is $7.25 an hour, 52 times more. DOTS. The idea is dead simple. While you're busy, it: → connects to your apps, over 4,000 to pick from → looks for ways to help on its own, with read-only access → sees a bug in Slack and starts investigating right away → takes a fresh design and turns it into a working app → notices you forgot to send an invoice, prepares it, and sends it once you say yes → drops what it finds into its own notes The honest part: The live demo at the keynote froze. The dot never answered. OpenAI hit it with 50,000 emails in a hacking test. One in three was an attack. Successful attacks: 0. But after a warning not to, it kept pushing in 17.4% of runs. For the first month it doesn't count against plan limits. What comes after, OpenAI hasn't said. 168 hours a week. 14 cents an hour. Right now everyone gets one dot. OpenAI already talks about teams of dots working together for you. In hours, one dot covers 4.2 full-time work weeks every week. Do the math on a team. Not joking. What's the first job you'd hand it?
This post quotes another post: x.com/mojunel/status/2103588484002852973