Triage inboxes and draft standard replies

An OpenClaw setup organized three inboxes, wrote everyday reply drafts, and reduced review time from 10 minutes to 2.

Done with OpenClaw· Intermediate· Mar 2026

What they did
A developer gave Donna, running through an OpenClaw gateway, access to three mailboxes through IMAP and Microsoft Graph. During business hours, every 30 minutes, it reviewed new mail, assigned each item to urgent, important, informational, or noise, and sent Telegram alerts for anything urgent. For regular messages, it produced draft replies and stored them instead of sending them, while also consulting a memory file with relationship notes.
How it went
Across 30 days, the triage decisions were right about 85% of the time, and no urgent message slipped through. The draft replies were about 70% ready to use, but tone, CC handling, cross-session context, and messages tied to relationships still caused trouble.
Tips
Set clear limits before turning it loose, especially for negotiations, legal topics, and ongoing relationships. In this case, the developer restricted drafting to routine mail and had the agent flag sensitive items for review instead of trying to handle them automatically.

Try it yourself with OpenClaw

Connect to my email accounts [account list] and review new messages from [time window]. Label each one as urgent, important, informational, or noise, then write reply drafts only for the straightforward ones that match [tone or style]. Do not send, delete, or submit anything on its own; stop first and ask me to approve any irreversible action. Finish with a brief recap of what needs my attention.

Fill in the [brackets]. Agents can make mistakes, so review anything before it is sent, paid or deleted.

Read the original on dev.to ↗

Get 3 agent uses like this every morning

With the day's top AI news. Free, and one click to unsubscribe.