No Standalone Email Address Yet: The Gap Between a Dot and a Coworker

We covered the mechanics in the previous article: a dot sends email through your account, under your identity, with every outgoing message checked against your Custom Rules. The structural fact underneath that mechanic is simple to state and genuinely consequential — a dot cannot yet have a standalone email address of its own, independent of yours. This article is about what that single gap rules out.

The coworker framing, and where it stops being literal

OpenAI's own launch materials described dots as reachable through ChatGPT, Codex, Slack, and Microsoft Teams, and leaned on language that positioned them as something close to an always-on employee — "agent coworkers" is the phrase that shows up in coverage of the launch. A real coworker, in any organization with a directory, has their own email address, their own calendar invites, their own identity that colleagues can address directly without going through you. A dot, as of late October 2026, has none of that. It operates as an extension of your identity, not as a separate one.

That's not a small implementation detail. It's the difference between "an agent that helps me do my job" and "an agent my team can treat as a member of it," and right now, per the inbox-mechanics guide covered in the previous article, dots sit firmly in the first category. A dot cannot receive an email addressed to it directly, because there's no address for anyone to send one to. It can't be CC'd by a colleague the way you'd CC a human teammate. It can't accumulate its own inbox history independent of your account.

What this rules out, concretely

A few things that would otherwise be natural extensions of the "agent coworker" pitch simply don't work yet as a result. Team hand-off — the idea that a dot could be assigned to a shared project and take direction from multiple people the way a human team member would — doesn't have the identity infrastructure to support it. Shared ownership of a dot across a team, rather than one dot per individual account, isn't a configuration the current email model allows for. And anything resembling external-facing use — a dot replying to a customer or vendor under its own identifiable name rather than yours — isn't on the table while it borrows your account for every send.

OpenAI has been explicit, per the same guide, that dots are positioned as personal agents rather than team-inbox agents: no shared ownership, no teammate hand-off, at least not yet. The missing standalone email address is the concrete technical fact that backs up that positioning — it's not just a stated product philosophy, it's a constraint baked into how the identity layer currently works.

"Agent coworker" is a marketing phrase that gets ahead of the plumbing. A coworker has an inbox. A dot, right now, has a permission to use yours.

The audit trail this creates, for better and worse

There's a secondary effect of borrowed identity worth naming, because it cuts in two directions at once. Since every dot action flows through your account, your Activity View becomes the single, complete record of everything that dot has done — there's no separate dot-owned log living somewhere else that you'd need to go check. That's a genuine simplicity benefit: one person, one account, one place to review.

The cost is that it also means a dot's actions are, from any outside system's perspective, indistinguishable from your own. A vendor, a colleague, or a compliance system receiving an email from your address has no built-in way to know whether you wrote it or your dot did, absent some sender-side disclosure that OpenAI hasn't described as a feature. For a personal productivity tool, that ambiguity is a minor wrinkle. For anything approaching regulated or audited correspondence — the kind of use case this series flagged as the real ceiling on this entire category — the inability to cryptographically or structurally distinguish "the human sent this" from "the dot sent this, with human approval" is a real limitation, not just a cosmetic one.

Whether this is temporary

Nothing in the available sourcing says whether a standalone identity for dots is coming later or is a deliberate, longer-term design choice tied to the personal-agent positioning. Given how much else about dots' roadmap — additional-dot pricing, specialist-dot details, what happens once the usage grace period ends — remains publicly unresolved three weeks after launch, it would be speculative to guess which way this goes. What's confirmed is the current state: a dot is a powerful extension of you, not yet a separate addressable entity your organization can route work to directly. Whether that's where dots stay, or just where they started, is one of the more interesting open questions in this whole series.

It's also worth noting that this constraint doesn't exist in isolation — it's part of a consistent pattern across every permission decision OpenAI has made public so far. The sandboxed cloud computer separate from your own machine at launch, the read-only scope of background proactive research, the per-message rather than per-approval Auto-Review checks on email, and now the borrowed rather than standalone identity, all point the same direction: OpenAI has consistently chosen the more conservative, more contained version of each capability over the more powerful, more autonomous one. That's a defensible strategy for a brand-new product category where the cost of a visible failure is high and the benefit of looking maximally capable on day one is comparatively low. It does mean that anyone picturing dots as further along the autonomy curve than they actually are, based on the "agent coworker" framing alone, is likely to be disappointed by the current reality rather than the eventual roadmap.

Part 36 of 47 in the ChatGPT Dots series. Previously: Your Dot and Your Inbox: What It Can and Still Cant Do With Email. Next: Four Settings, One Dot: How Custom Rules Actually Gate an Action. Browse more builder insights or explore AI skills at aiskill.market.