ChatGPT Dots
47 articles in series
Forty-seven articles on OpenAI's Dots — the always-on GPT-6 Astra agents announced at DevDay 2026, each with its own cloud computer and browser, wired into 4,000+ apps through ChatGPT Space. Mechanics, rollout, use cases, the first month's scorecard, and honest scrutiny of the permission model, pricing gaps, and how Dots stacks up against Grok Bot and Meta Muse.
The First Month of Dots: A Scorecard Against What OpenAI Promised
OpenAI made specific claims at dots' launch. One month later, here's what held up, what's still unverified, and what OpenAI still hasn't addressed.
The Grace Period Ends: What Happens When Dots Usage Starts Counting
OpenAI gave dots a free month where usage didn't touch your plan limits. That window is closing in late October, and OpenAI still hasn't said what comes next.
Three Weeks In: What OpenAI Still Hasnt Told Dots Users
Dots launched three weeks ago with a grace period, a narrow rollout, and several open questions. Most of them are still open — here's the running list.
Why a Dot Isnt a Team Inbox Agent, and What That Rules Out
Dots are built as personal agents, not shared team tools. No ownership hand-off, no multi-person direction. Here's what that design choice actually costs and buys.
Your Dot and Your Inbox: What It Can and Still Cant Do With Email
A dot can draft and check an email against your rules before sending — but it still borrows your inbox rather than having one of its own. Here's the actual mechanics.
Agent 365 and the Enterprise Path Microsoft Is Building for Dots
OpenAI is routing enterprise Dots through Microsoft's Agent 365 security controls, treating agents as machine identities with onboarding, credentials, and offboarding. Here's what that means.
Auto-Review: The Safety Net Between a Dot and a Consequential Action
Before a dot executes something consequential, Auto-Review checks it against your instructions, your Custom Rules, and OpenAI's safety baseline. How much does it catch?
ChatGPT Space, Explained: Where Dots and Humans Meet
OpenAI replaced Library with Space, a shared workspace where employees, ChatGPT, Codex, and dots all work against the same context. Here's what actually changed.
Custom Rules: How Dots Learn What They Are Allowed to Do
Custom Rules let you pre-approve, block, or require sign-off on categories of action before a dot ever attempts them. Here's how the mechanism actually works.
Data, Training, and Dots: What OpenAI Will and Will Not Use
OpenAI drew specific lines around what a dot's background research and private notes feed into model training. Here's what's excluded, what's opt-in, and what's still unclear.
Dots and the Bubbly Avatar: Why Cute Branding Matters
OpenAI gave its always-on agents a cartoonish, floating-bubble look almost identical in spirit to Meta's Muse. That's not an accident — it's a deliberate answer to a trust problem.
Dots vs Codex: Two Very Different Kinds of Autonomy
Both are OpenAI agents that can write and ship code. But a dot and a Codex session are built on opposite assumptions about how long an agent should keep running.