Most of the internet still expects a human with an inbox at the door.
Giving an agent its own email closes the loop on sign ups, codes and receipts. Why that plumbing lets agents take whole jobs, and why sends still need approval.

Grok Bot now has its own email. Bots can sign up for services, contact businesses and schedule time, using an address of their own. It's rolling out now, and on teams an admin has to switch it on.
This is where it gets interesting.
So much agent work used to stall at the same step. Create an account, wait for the verification code, confirm the receipt. Most of those needed a human to stop and copy a code across. An agent with its own inbox can finish that loop by itself.
Sign-ups, codes, receipts and reports are the plumbing of real work. When an agent can handle the plumbing, it can take on whole jobs instead of the middle of a job.
My agents claimed one inbox on launch day. It's where sign-ups, verification codes, receipts and reports land now, all in one place I can check. In my setup, sends still need my approval. Inbound mail gets the same care. Email can carry prompt injection, so my agents treat every message as data, not instructions. They never click links or share codes because an email asks, and anything that looks like phishing gets flagged to me.
That's the builder lesson. Giving an agent an identity is the easy part. Deciding what it's allowed to say, and what it's allowed to believe, is the design work.
Give your agents a front door. Keep the key to the outbox.