Building a permission layer for AI agents

Would you let an AI handle your invoices and orders if you could set limits and approve anything unusual from your phone as a business owner?

I've been testing AI agents / workers that handle repetitive admin work such as reading supplier invoices, flagging low stock, drafting reorders, answering routine customer messages etc...

The agents work but the problem is trust. No owner hands an AI the keys to their bank account or their customer database example like WhatsApp, because if it makes a mistake it makes it confidently and fast.

So I'm thinking of building a version which works like hiring a junior employee with strict rules.

• It can act alone only under limits you set (examples: payments under $150, only to suppliers you've approved)
• Anything bigger or unusual you get a message on your phone to approve or reject, one tap.
• Every single thing it does or tries to do is written in a log you can read in plain language.
• One button shuts it off instantly

Its not "trust the AI" it's "the AI physically cannot exceed the authority you give it."

Questions for people running a business:

1. Would you use something like this, or is any AI near your money a hard no?
2. What's the first task you'd hand over invoices, reordering, customer replies, something else?
3. What limit setup would make you comfortable?
4. What would this need to save you time or money to be worth paying for as a subscription.

Trying to find out if this solves a real problem or just an imaginary one.