Top of Permission — Agentic Commerce (3 of 4)
Issuers face a fork: be the passive endpoint of an agent's decision, or the layer that governs it.
Third of four from the agentic commerce research. This one's about where the real value sits.
When an AI agent goes to pay, something has to decide whether that specific transaction is allowed. Not the card number — the permission behind it. Is this agent authorized to act for this cardholder? For this kind of purchase? Up to this amount? Under these conditions? Right now, at this moment, given everything it's already done today?
That decision has to live somewhere. And it's a genuinely new and valuable position in the stack — one that, right now, is largely unclaimed. I've been calling it the top of permission: the governing layer that sits above the transaction and answers the authorization question before a charge is ever attempted.
Here's the fork I think every issuer faces. One path is to remain the passive endpoint — the agent and the networks make the decisions, and the card simply settles whatever arrives. The economics of that path are thin and getting thinner; you become a balance sheet that other people's decisions route through, competing on cost. The other path is to become the governance layer itself: to own the rules for what an agent may do with your credential, to hold the record of what the cardholder actually consented to, to set and enforce the limits, to handle revocation cleanly, and to be the place where the permission question gets answered.
The second path is harder, and it's where the durable value is. It also raises the bar operationally, and I want to be precise about this: real-time, granular, auditable permissioning simply isn't what card authorization systems were originally built to do. That's a new requirement — and it's new for every processing environment, regardless of vintage or provider. The point isn't who's incumbent. The point is who builds the capability.
That's the honest framing, because this is where I think the strategic mistake gets made. It's tempting to treat agentic permissioning as a message to pass through — one more field to accept and forward. But the institutions that treat it as a product to own, something they build and stand behind, position themselves at the top of a stack that is still forming. The ones that wait will inherit whatever standard someone else sets, and then spend years conforming to it.
Endpoint, or governance layer. It's a choice available right now, while the layer is still unclaimed. It won't stay unclaimed for long.
The full argument is on The Payments Corner.
Franco Di Pietro
The Payments Corner
30+ years across payments, fintech, banking, and financial infrastructure. Operator-level perspectives on the systems that move money.
Related Insights
Citi's card-linked offer acquisition bets on cost, not on proven category economics
Citi's acquisition of Kard Financial rests on operational and financial logic—lower rewards spend, faster platform build—but the card-linked offer category itself has never demonstrated standalone profitability. Cardlytics, the only pure-play, has burned $1.15bn cumulatively since its 2018 IPO and generated negative shareholder equity last year.
The PayPal Valuation Paradox
Stripe and Advent walked away, PayPal lost nearly 13% in a day, and the takeover premium disappeared. The failed deal exposed a more important question: how much value is the public market actually assigning to Venmo?
Kraken's card is a bet on controlling the funding decision, not the transaction
Kraken's debit card—with its 2% rewards across multiple asset balances—is not a payments product disguised as one. It is a wallet-stickiness play that inverts traditional card architecture: instead of a card accessing a single account, the platform now decides which of many balances funds each purchase, moving the locus of control from the network to the issuer's financial operating system.