OpenAI Dots: How Autonomous Workplace Agents Actually Work
Alexander Embiricos explains how OpenAI Dots give workplace AI agents separate computers, dedicated Slack handles, and on-the-fly HTML interfaces.
🔴 LIVE from OpenAI DevDay
With John Coogan, Jordi Hays, Sam Altman, Alexander Embiricos, Andrew Ambrosino, Peter Steinberger, Tibo Sottiaux, Kath Corvac, Casey Neistat · Sunday, October 4, 2026
Broadcasting live from OpenAI DevDay 2026, John Coogan and Jordi Hays interview key OpenAI leaders, researchers, and creators about major announcements including Dots, Codex Cloud, ChatGPT Spaces, and custom hardware. Guests Sam Altman, Alexander Embiricos, Andrew Ambrosino, Peter Steinberger, Tibo Sottiaux, Kath Corvac, and Casey Neistat explore the shift toward proactive autonomous agents, computer use capabilities, developer ecosystems, and the cultural balance between digital automation and constrained physical devices.
Key takeaways
- OpenAI Dots run on isolated cloud computers, granting agents the operational range of a human worker rather than restricting them to basic API wrappers. Read more →
- OpenAI rewrote ChatGPT.com in five weeks to position it as an executive app capable of handling agent workflows across desktop and cloud. Read more →
- Mod Retro built a custom plugin for OpenAI Codex that lets developers vibe-code games and flash them directly onto physical cartridges. Read more →
- Peter Steinberger built OpenClaw to push models beyond passive chat and give them direct execution tools to buy goods, install local software, and manage desktop workflows. Read more →
- OpenAI plans to bring its in-house custom chip program online in the first half of 2027 to manage surging inference demand. Read more →
- Computer use ran ten times faster year-over-year by cutting reasoning overhead, parallelizing execution steps, and speeding up safety checks. Read more →
6 articles from this episode
Andrew Ambrosino on Fast Models and Dynamic UI Generation
Andrew Ambrosino explains why ultra-fast model inference and local machine access rewrite how teams build agent software.
Casey Neistat on Mod Retro, Codex Cartridges, and AI Art
Casey Neistat explains why physical hardware constraints and human taste matter more than ever in an era of infinite AI software.
Peter Steinberger on OpenClaw and the Agent Wars
OpenClaw founder Peter Steinberger explains why user-side AI agents will bypass corporate walled gardens and rewrite the operating system.
Sam Altman on OpenAI's 2027 Chips and AI Alignment
Sam Altman breaks down OpenAI's 2027 custom chip timeline, the death of model pickers, and why AI alignment remains an unsolved science.
OpenAI's Tibo Sottiaux on Agent Speed, Scaffolding, and Guardian Models
Tibo Sottiaux explains how OpenAI achieved a 10x speedup in computer use and why deleting outer code is essential for smarter models.
More TBPN episodes
- Some Personal Agent News, Mongo Got Zucked, SNL, Gstaad Guy In The UltradomeOct 4 · 7 articles
- Cambridge AI Safety Plan, Instinct CEO on Invest Like the Best, Zuck Hires MongoDB CEO | Diet TBPNOct 4 · 4 articles
- Harvey’s Margins, Are Insects Worth More Than Humans?, Meta Tests Human Help for Muse | Diet TBPNSep 27 · 5 articles
- Jensen’s Ezra Interview, EV News, Nathan Fielder’s Elizabeth Holmes Doc | Diet TBPNSep 27 · 3 articles