Consent in agentic payments

An agent that can browse, compare and transact is not far from an agent that pays. The technical path is well trodden; the harder question for institutions is what a customer's consent has to mean once someone other than the customer is the one pressing send.

Where consent gets harder

Delegation is easy to describe and difficult to bound. A customer who says "renew my subscriptions under two hundred dollars" has given an instruction that is reasonable in spirit and ambiguous in practice. Does it cover the tier that changed price? The vendor the agent found rather than the one expected? The renewal two weeks early?

• Scope — how wide a mandate is, and whether the agent's action fell inside it.
• Timing — whether confirmation was given for this transaction or for a class of transactions.
• Awareness — what the customer was actually shown before authorizing.
• Revocation — how a mandate is narrowed or withdrawn once it is live.

Questions institutions are raising

• What exactly did the customer authorize, in words they would recognize?
• How is the boundary between authorized and not authorized captured, rather than inferred after the fact?
• What evidence survives an examination, and who can verify it independently?
• How is liability allocated when an agent acts inside a mandate the customer believes they withdrew?

None of these are reasons to move slowly. They are reasons to decide, in advance, what a consent record has to contain — before the volume arrives and the question is asked retrospectively.

A practical starting point

Most institutions do not need to wait for a general framework. They can pick one narrow, high-value category — a transfer ceiling, a renewal class, a disbursement type — and make the authorization record for it explicit: what the customer saw, what they confirmed, when, and who attested it. That single record answers more questions than any policy document about agent use.

Notaries have verified what's real for centuries. We're bringing that same trusted concept to the digital world.

ReLeaf Financial's Proof of Intent™ is intended to fit into that pattern: an SDK that captures authorization at the point of decision and returns a signed record to the system that has to rely on it. The relying institution keeps control of what happens next — the evidence simply makes the decision defensible afterward.