Can Claude Cowork Send Emails? The Honest Answer (2026)
Yes on both mailboxes now, with an approval step, and with one condition on the Microsoft side. On Gmail, Anthropic’s connector listing carries a Read & write badge and states that Claude “can also draft, send, reply to, and forward emails, and asks for your approval by default before anything goes out.” That is a change: the Google Workspace connector was draft-only for most of 2026, and a great deal of writing about it, ours included until this update, still says so. On Outlook, sending runs through the Microsoft 365 connector’s write tools, which are off by default and switched on by a tenant administrator.
There is a third fact that matters more than either, and it is the one people miss when they picture Cowork handling their email: nothing arrives at Cowork. It has no inbox of its own, so even where sending works, the sending starts with you.
Why the answer depends on which mailbox
Cowork does not have its own email connectors. It uses the same ones as Claude, so its email capability is exactly Claude’s email capability plus the ability to run for a long time on a schedule.
Gmail, through the Google Workspace connector. Sends, replies and forwards. Anthropic’s help documentation lists all three as capabilities and notes that Claude asks for your approval before each one by default. No administrator is involved, which makes this the lower-friction of the two paths.
Outlook, through the Microsoft 365 connector. Read-only until July 2026, when Anthropic added write tools covering email, calendar, and files. Those tools can send email. Two conditions: an administrator has to enable them, since they are off by default, and on a managed work tenant that is not your decision.
So “can Cowork send email” resolves to: yes on Gmail with your approval, and yes on Outlook if IT says so.
The structural limit: it has no inbox
This is the one worth sitting with, because it is not a setting anyone can change.
Cowork is something you prompt. You open Claude, describe an outcome, and it works. Even a scheduled task is a prompt you wrote earlier running on a cadence you chose. There is no email address a client, a lead, or a colleague can write to and get a response from. Nobody outside your Claude account can hand it work.
If the trigger for the task is “someone emailed me,” Cowork has no native way to know that happened. You notice, then you go and prompt it.
That rules out the thing most people are actually asking about when they search this. An enquiry landing at 11pm does not get an answer at 11pm. It gets an answer when you next open Claude and mention it.
What Cowork genuinely does well with email
Worth being fair about, because the gaps above do not make it useless:
- Digest and triage. A scheduled task that reads your inbox and produces a summary of what needs you, running remotely so nothing of yours has to be switched on.
- Long drafting runs. Working through a backlog of replies in one pass, with the context of attached documents and connected tools.
- Multi-step work that ends in an email. Research, build the deck, then send the message that carries it. Both paths can now finish that last step, Gmail with your approval and Outlook once write tools are on.
- Attachment comprehension. Reading
.docx,.xlsx,.pptx, and.pdffiles in a thread without downloading anything.
Two more limits to price in
It is metered, and this work burns the meter. Cowork is bundled into your Claude plan rather than sold flat, and every session draws down your allocation. Long multi-step runs are exactly the token-intensive kind. A daily inbox sweep is a fine idea until you see what a month of it costs against your window. See Claude Cowork limitations.
A scheduled run will not send from Outlook. Claude’s Microsoft 365 write tools need a live conversation, so an unattended task can read and prepare and then waits for you. The Gmail path does not carry that restriction.
Scheduled is not the same as triggered. Tasks fire on a clock. Set one for 8am and a message arriving at 9:15am waits until tomorrow. For a digest that is fine. For anything time-sensitive it is the wrong shape.
Setting up what does work
For an Outlook mailbox where sending is the goal:
- Open Settings → Connectors in the Claude app or on claude.ai.
- Connect Microsoft 365 and sign in with your work account.
- Ask your administrator to enable write tools on the connector. They are off by default, and there is no way around this from your side.
- Test with a message to yourself before letting it write to anyone else.
For Gmail, connect Google Workspace the same way. Sending needs no administrator, and each send asks you first unless you move that tool to Always allow under Tool permissions.
If you want email that answers on its own
Carly AI is built around the gap described above. Each agent gets its own name, email address, and memory, so people write to it directly and it replies and does the work underneath. That is the reachability a prompt-driven agent structurally cannot offer.
It fires the moment mail arrives rather than on a clock, sends real email with attachments from the address that received the thread, and does it on both Gmail and Outlook without an administrator switching anything on. Several mailboxes are held at once, each authorized separately with a role you name, so a work account and a personal one both work and every action reports which one it came from.
The practical difference: an enquiry at 11pm gets a real answer at 11pm, and the follow-up goes on the calendar. It runs in the cloud on a $35/month plan, with free Zapier-style workflows underneath for the steps that do not need a model.
Where Cowork is better: long interactive analysis, building documents and decks, and driving a browser. They solve different problems and a lot of people run both.
FAQ
Can Claude Cowork send emails from Gmail?
Yes. The Google Workspace connector drafts, sends, replies and forwards, and asks for your approval before anything goes out by default. Guidance describing it as draft-only is out of date.
Can Claude Cowork send emails from Outlook?
Yes, through the Microsoft 365 connector’s write tools added in July 2026. They are off by default and a tenant administrator has to enable them first.
Does Claude Cowork have its own email address?
No. It has no inbox and nobody outside your Claude account can send it work. Every task starts with you prompting it, including scheduled ones you set up earlier.
Can Cowork reply to an email as soon as it arrives?
No. There are no inbox triggers. Scheduled tasks run on a fixed clock, so a message arriving between runs waits. Carly fires on the message itself.
Is Claude Cowork email included in the free plan?
No. Cowork is not part of the free Claude plan at all. It comes with Pro and above, and every session draws down that plan’s usage allocation.
Related: Claude Cowork limitations · Can Claude send emails? · Can Claude send emails from Outlook? · Claude Cowork alternatives · Best AI email assistants for Outlook
Ready to automate your busywork?
Carly schedules, researches, and briefs you—so you can focus on what matters.
See what people say
"Before Carly, I relied on a Calendly link, but the whole process felt impersonal and not very professional. Carly changed that by handling all the back-and-forth, so I'm no longer stuck in endless email threads trying to line up schedules.
Now Carly reaches out to candidates, shares my real-time availability, lets them pick a slot, then sends a Zoom link and drops it straight into my calendar. She sends reminders to both of us before each call, which has significantly reduced no-shows and last-minute confusion.
On top of scheduling, Carly acts like a full executive assistant, sending me my schedule the night before so I can prepare for each call. It reminds me of the old x.ai assistant, but Carly is noticeably smarter, faster, and better suited to my healthcare recruitment business."
