Dots gets another round of updates: Codex work from chat context, editable Scheduled Tasks, and a long fixes list

#What happened

OpenAI posted a new round of Dots updates on October 13, 2026. Dots can now start Codex work and follow up on threads using context from your chats, Codex threads, and automations. Users can now review and edit Scheduled Tasks. The round also includes creating a dot on iOS and Android, then opening ChatGPT directly into it.

The update also ships a long list of fixes: cloud browsing and the sidebar should load faster, with smoother scrolling, fewer pauses, and less network usage. Users should receive fewer phone notifications while actively using ChatGPT. Code blocks and attachments are cleaner, character rendering in Safari and Firefox has been fixed, and threads are reused and renamed more intelligently. Users should also see fewer repeated "created task" messages. Plans are clearer, activity that fails to load will retry, safety banners are more stable, and approval buttons should no longer be clipped or hidden. The update also includes fixes for Enterprise web and voice access.

#Why it matters

The update pushes Dots further toward continuous, context-aware work: a dot can now pick up Codex work from what you already discussed instead of starting cold, and Scheduled Tasks become something users can inspect and change rather than set and forget. The fixes target the everyday friction of an always-on agent: faster browsing, fewer duplicate notifications, and approval buttons that are actually visible when a dot needs a decision.