TECHNOMALIST
Anthropic has expanded Claude's Gmail integration so the assistant can now send, reply to, and forward messages without stopping for manual confirmation on each one. The change builds on an existing Google Workspace connector that already gave Claude access to Gmail, Google Calendar, and Google Drive. Until now, the missing action was actually sending an email; users could have Claude draft a message but had to hit send themselves. With this update, that final step can be automated.
According to Digital Trends, the new behavior is available only on paid Claude plans. Users can ask Claude to respond to an email thread, and the AI will draft the reply and send it. When the appropriate setting is enabled, Claude does not pause to wait for approval before each outgoing message. Approval remains the default. Individuals can decide whether Claude needs confirmation before sending, while Team and Enterprise administrators can set rules for whether members can turn off repeated approval. This means the degree of autonomy is configurable.
The update also lets Claude forward emails, which broadens the integration beyond simple replies. That makes the assistant operate inside the inbox more like an agent that can act on a user's behalf, rather than a tool that only summarizes threads or suggests replies. The distinction matters because writing an email and sending one are fundamentally different. A chatbot can produce a perfectly worded reply and leave the final decision to the human. An agent that can press send crosses into taking action.
For people who handle repetitive email, the change could save significant time. Routine follow-ups, replies to common questions, and forwarding messages are tasks that an AI agent can handle with minimal human involvement. Instead of reviewing every draft, a user could set the policy once and let Claude work through the inbox.
The automation is not without risk. If Claude misunderstands context or tone, an incorrect draft is easy to fix before sending. Once a message is sent, it cannot be taken back. Anthropic's approval system acts as a safety valve, but disabling that checkpoint places more responsibility on users and, for workplace accounts, administrators. That trade-off is central to the feature's design: convenience on one side, control on the other.
Beyond the immediate product update, the change points to where AI assistants are heading. The goal is increasingly to answer questions and also to interact with the services people already use, completing tasks inside them rather than only offering suggestions. Claude's Gmail capabilities are limited to paid plans, and approval settings vary according to account type. For users willing to grant more autonomy, email could become one of the first everyday tasks where an AI moves from suggesting an action to actually carrying it out.
See an error? Read our corrections policy or email [email protected].
