Claude Duet: Approval Checks Before an AI Agent Sends
Changed attachments, changed recipients, revoked permission and uncertain sends: four proposed checks for keeping the decision visible and bringing back the receipt.
The short answer
Approval should follow the final action package: recipient, message and attachment. If any of those change, ask for a fresh decision; if a send times out, check what happened before retrying. This episode proposes acceptance checks inspired by Claude Duet, rather than reporting a product test.
Public-source analysis and proposed acceptance checks. We have not tested Claude Duet.
Check the final recipient and attachment before action. If a send times out, inspect its result before retrying.
Questions this video answers
What should an AI agent show before sending an email?
The proposed workflow shows the final recipient, message and attachment, asks for a decision on that exact package, and rechecks permission before sending.
What should happen if the attachment or recipient changes?
The earlier approval should be reconsidered because the destination or information leaving the business has changed.
Does this video show that Claude Duet failed these checks?
No. The video presents public-source analysis and proposed acceptance checks. We have not run those checks against Duet.
Full video transcript
Read the transcript
I approved the email. Then I changed the attachment. Which version should the agent send?
Here’s my rule. Show me the final package. Get my decision on that version. Then bring back the receipt.
Claude Duet got me thinking about this. Anthropic says it can prepare work in the background and ask before important actions. That opens up real possibilities for business owners.
We turned that announcement into a case study. This is public-source analysis, not a product test.
Let’s walk through an example. An agent drafts a client email. I approve it. Then I change the recipient and swap the attachment.
Watch those two fields. The person receiving the information changed. The information leaving the business changed. My earlier approval should not silently carry over.
The proposed workflow is simple. Prepare the final action. Show the recipient, message and attachment. Ask for a decision on that exact package. Recheck permission before sending. Then verify what actually happened.
Our case study proposes four checks.
Change the draft or attachment. Does the workflow pause for a fresh decision?
Change the recipient. Does the approval show the final destination?
Revoke permission. Does the action stop and explain why?
Let the send time out. Does the system check the first attempt before trying again?
That last one matters. A timeout does not tell you whether the email was sent.
For the receipt, I want the final recipient, the message identifier and the time. Enough to connect the action I approved with the result we can actually verify.
Those are proposed acceptance checks. We haven’t run them against Duet, and we’re not claiming the product failed them.
That’s the focus behind Whitewater AIOS and the WWAIOS framework. Make the evidence, permissions and decision behind an AI action visible and defensible. Reduce distortion between reality and decision.
Take the four checks in the case study and apply them to one workflow you’re considering. Keep the decision visible. Bring back the receipt.