Draft Replies in Your Voice, Not Cowork's

You ask Cowork to reply to a client who pushed your deadline. The draft comes back polished: polite, complete, every comma in place. It says the right things, so you send it.

Two days later a colleague on the thread pings you: "Did you write that, or did a bot?" You read it again. It opens with "I hope this message finds you well." It "wanted to circle back." Nothing is wrong. Everything is generic. It sounds like every other AI-drafted email your recipient got that week, because that's what it is.

The tool wasn't the problem. You asked for a reply and never told Cowork how you sound, so it defaulted to the blandest safe register it knows.

The fix is small: show it your voice once, and reuse it.

#Why the default voice betrays you

Left to itself, Cowork writes the average of everything it has read. That average is competent and faceless: long greetings, hedged transitions, a tidy sign-off. It reads as careful, which to a real recipient reads as distant, or worse, automated.