Lesson 6 of 7

Keep it on a leash

6 min read

Your assistant can now send emails and spend money on your behalf. That's powerful — and a little terrifying. How do you get the help without handing over the keys?

Gate what it can't undo

The fix is a leash: clear limits on what the assistant may do alone. Low-risk actions — searching, reading, drafting — can run freely. But anything that spends money, sends a message as you, or can't be undone stops at a checkpoint and waits for your OK. This is [human-in-the-loop](glossary://human-in-the-loop): the assistant proposes, you approve.

Human-in-the-loop means the assistant never takes a risky, irreversible action alone — a person signs off first.

Permissions set the fence

Alongside the checkpoint sit permissions: what each tool is even allowed to touch. Read-only versus read-and-send. Your calendar but not your bank. Set the fence narrow and the assistant works safely inside it — and for the truly risky edges, you still hold the final yes.

The goal isn't a fully autonomous assistant — it's a useful one you can trust, because the risky actions are fenced and gated.

Be most careful with actions that move money, send messages, or delete things — the ones you can't take back. Read what the assistant proposes before you approve; a fast yes is how automation goes wrong.