What is an Approval Gate?

What it is: A pause built into an AI workflow where the AI stops and waits for you to say “yes” before doing something that cannot be easily undone — sending an email, publishing a post, deleting a file.
Who it’s for: Anyone using AI to help with tasks that have real-world consequences: publishing, communicating, modifying data.
Best if: You want the speed of AI assistance without the risk of the AI acting on a misunderstanding.
Skip if: Your AI is only doing read-only tasks where nothing can go wrong.

The problem with powerful AI isn’t that it does nothing. It’s that it does too much, too fast, too confidently.

An approval gate is the solution. It’s a checkpoint in the workflow — a moment where the AI has figured out what it wants to do, written the action out, and then stops. It shows you what it’s about to do and asks: do you want me to proceed?

Think of the “Are you sure?” dialog that pops up when you empty the trash on your computer. That’s an approval gate. The system isn’t preventing you from deleting things — it’s just making sure you meant it before it does something that can’t be undone.

In AI workflows, approval gates typically protect:

• Sends — emails, Slack messages, newsletter publishes
• Writes — overwriting a file, updating a database record
• Deletes — removing content, canceling subscriptions, clearing data
• Publishes — pushing a blog post live, posting to social media

The flip side: approval gates for everything defeats the purpose of having AI help. A well-designed system lets the AI read and research freely — those actions carry no risk — while gating anything that leaves a mark on the world.

Some systems have a time limit baked in: if you don’t respond within five minutes, the pending action is automatically cancelled. That way a risky action can’t sit unreviewed while you’re away from your desk.

Related

• What are Hooks? (in Claude Code)
• What is Claude Code?