Hej product hunt! I'm douwe, the technical founder of attesso. The problem: as AI agents start spending money on behalf of people, "the agent did it" isn't a good enough answer. There's no neutral layer that says who authorized a purchase, what exactly they approved, and proof the decision stayed within those bounds. Attesso is that layer, authorization and proof for AI-agent purchases. You give your agent a bounded mandate, your end user approves it once with a passkey, and attesso returns a deterministic ALLOW / DENY for each proposed action, plus signed evidence you can verify independently. The person doesn't have to be present at checkout, the approval is already in place before the task starts, with the action and timing bound to it. It's provider-neutral: attesso never moves money, holds cards, or executes the action. Your agent does, against your own providers. I'd love to hear from the builders here: where would you require an explicit, human-approved boundary before an agent acts?

Attesso

9 lines of code between your agent and a payment.

9 lines of code between your agent and a payment.