Israel just published a draft stablecoin issuance law. It leans heavily on the GENIUS Act and MiCA. Same disclosure requirements, same reserve rules, same redemption protections. Which is exactly the problem.
Both GENIUS Act and MiCA were written for a holder who reads disclosure documents, weighs risk, and makes an informed decision before buying or redeeming. That holder is disappearing.
An autonomous agent doesn't read disclosures. It opens a wallet, trades, and redeems thousands of times a day, optimizing for cost and speed, not comfort or trust in a brand. Every framework built so far assumes a human is still making that call.
This isn't a criticism of Israel for copying international standards. Relying on a proven model, rather than reinventing the wheel, is the right instinct. Payments regulation did the same thing for years with PSD.
The real question is different: are the assumptions inside GENIUS Act and MiCA themselves still valid. If they are not, importing them just copies the same blind spot into a new jurisdiction.
Interestingly, the Israeli draft already builds half the bridge without meaning to. A ban on paying interest, a ban on redemption fees, recognition of foreign issuers. That is exactly the infrastructure a payment focused, non speculative instrument needs to serve autonomous agents.
The open question is who builds the other half. Not just in Israel, but everywhere still regulating for a holder who reads paperwork.
What would regulation written for agents actually require. Machine readable disclosure. Authorization frameworks. Redemption speed standards. Would love to hear your thoughts.
#agenticpayments #stablecoins
Would you hand ChatGPT your credit card?
Most people say no instantly. But ask yourself what is actually different between that and saving your card in Apple Pay. You already do not see the tokenization. You already do not control the process. You just trust the system.
What changes with agentic payments is not the technology. It is the feeling of control. Today we have an illusion of presence: I saw the screen, I clicked the button, I was there. With autonomous agents, that illusion disappears. You are not approving a transaction anymore. You are issuing a mandate.
That is why this race will not be won by whoever builds the smartest agent. It will be won by whoever builds systemic trust. And nobody has been doing that longer than the card schemes. Decades of authentication, liability frameworks, tokenization and risk management, built so we never had to understand what happens behind the screen.
Agentic payments actually need two separate layers of trust. One: that the agent is authorized and is who it claims to be. Two: trust in the payment itself, that the funds are safe and the same protections we have today still apply if something goes wrong.
Schemes are building the second layer. The first one, identity and authorization, is still wide open, and the gap is being filled in very different ways.
A Y Combinator backed startup called Agentcard took the most cautious answer possible. Single-use virtual cards, built on Visa rails, that self-destruct after one transaction. Every charge requires your real time approval. No autonomy, no standing mandate, just a human in the loop on every single purchase. It works precisely because it does not trust the agent at all.
Mastercard is betting on the opposite end with Agent Pay for Machines, designed for continuous, high frequency machine to machine payments with no human in sight. Visa is pushing its own agents deeper into ChatGPT with Agent Scoring and an Agentic Registry.
One camp is building infrastructure for full delegation. The other is building infrastructure for permanent supervision.
The real question is not whether the technology works. It is how much control you are actually willing to give up, and whether you even get to choose.
I've spent 20+ years in payments- as a lawyer, as EVP Strategy at one of Israel's largest card companies, and as founder & CEO of Hyp, which today processes 40%+ of all card transactions in Israel.
I'm starting to write here about agentic payments, strategy, and where this industry is actually heading.
First take:
Everyone assumes AI agents will kill credit cards. I'm not sure that's right.
Think like an agent for a second. You need to pay a supplier in the US from Israel. A2A? Doesn't work -- no global infrastructure. Visa/Mastercard? Works in seconds.
Cards win. No contest.
But domestic transactions? Germany to Germany, US to US? Open Banking already works there, and the cost difference is sometimes 10x in favor of A2A.
So what we're likely heading toward is a split: cards dominate cross-border, A2A quietly erodes domestic market share.
And here's what makes it interesting: most transactions globally are domestic.
The real battlefield for agentic payments isn't where everyone is looking, and we haven't even started talking about new railsץ
@MikeE_3_14 אני ממש אוהב את החידות שלך, וממליץ שכאשר אנחנו קמים ליום קשה, החידה תהיה ממש קלה, כמו היום, כדי לתת תחושת מסוגלות. בקיצור ליניאריות בין קושי השאלה למצב רוח הלאומי....
@Eran_Efrat Search fund קלאסי. מדהים
אגב, מנסיוני, התהליך של ה PMI (כלומר, אחרי הרכישה) הוא החלק הקשה ביותר, ולאו דווקא בגלל המוצר, אלא בגלל הצורך לשנות DNA . בהצלחה, אחלה מהלך!
@ItaiBoublil כל הכבוד. על האומץ. על הדרך. על הפוסט. ולא לוותר. להיות יזם ומייסד זה אירוע שברוב הזמן הוא לא מתגמל..אבל יש סיבה שאנחנו בוחרים בו. בהצלחה!