AI emails in Superhuman are now noticeably more useful

I’d like to share a quick update from the world of AI and email: Superhuman has upgraded auto-draft, making AI-generated drafts feel much closer to natural human communication.

What has changed:
— the system now identifies important emails on its own and suggests a reply;
— drafts reflect my tone based on past conversations;
— 2 additional reply versions are available;
— the model learns from my edits and becomes more accurate over time.

In tests, some of these drafts were sent with almost no editing. That said, the errors remain: sometimes the AI is overly optimistic or suggests an inconvenient meeting time.

Why it matters: this approach can save real time when inboxes are full of repetitive replies, but human oversight is still essential.

Would you trust AI drafts in your inbox?

#AI #productivity #email #HackerNews