THE HOOK
OpenAI just gave ChatGPT a coworker that never clocks out.
THE GIST
OpenAI is rolling out Dots, always-on GPT-6 Astra agents with dedicated cloud computers and plugins into 4,000+ apps via ChatGPT, Slack, or Teams. Sensitive actions need explicit consent. Rollout starts on Pro, Business Premium, and Enterprise in eligible markets.
THE TAKE
This is the product bet after a summer of rogue-agent headlines: ship the always-on teammate anyway, wrap it in consent gates, and let Pro users normalize agent work before regulators finish drafting. Isolated cloud computers and approval prompts are table stakes, not a safety thesis. If Dots actually invoice, debug, and ship while you sleep, every other chatbot suddenly looks like a toy.
OPINION
OpenAI's own blog pitches Dots as reclaiming attention while the industry still digests agent sandbox escapes. Watch whether enterprise specialist dots and access-management features ship as fast as the consumer pitch — that's where liability and identity get real.
- Story type: Breaking news
- Impact: High
- Confidence: High (official announcement)
- Audience: Developers
- What changed: OpenAI began rolling out always-on Dots agents with cloud computers and multi-app plugins on GPT-6 Astra
- What to watch: Enterprise specialist dots, consent UX under load, and whether agent incidents force a pause
- Reading time: 1 minute
