When an AI Agent Acts on Your Behalf, Who Says Yes?

I was having dinner with some friends this week, and when the topic of AI agents came up, I jumped at the chance to share what I've been up to with Aiyo.

During the evening, there were several moments when I found myself proud of my ability to make Aiyo relatable and understandable, and even though this isn't an exact transcript of our conversation, it is pretty close to it.

I'm sharing this because not everyone is a developer, but when we are dealing with emerging technologies and economies that will affect every day individuals, we also need messaging that's intentionally created to inform and empower them, while using language that's familiar.

So I invite everyone, from the most brilliant tech minds to someone completely against the rise of AI and autonomous systems, to continue reading, to learn more about both the importance of Aiyo and our goals for making the future of autonomous systems and AI agents a safer and more trusting one.

---

So Ray, you're working in AI or something, man? *laughs*

Yeah, I actually started a company. It's called Aiyo. I learned about AI agents being able to buy things on their own, and I was like, um, they aren't about to be spending my money!

Nah, for real, bro! This AI stuff is getting out of hand!

Seriously! We're moving way too fast with it, and a bunch of companies and sectors, including banks and healthcare, are switching over our money and data to these things.

…and we're just somehow supposed to be okay with this?

We're supposed to trust the AI agents at the same companies who have had data leaks and breaches in the past? These are the same companies who we've had to take to court for these same matters, and now, suddenly because they want to use AI agents and autonomous systems, we are supposed to blindly trust their new shiny technology to move and access our money and data all by themselves? No. That's not cool.

Yo, I swear I just heard about some agents doing a cyberattack this summer.

That was the OpenAI and Hugging Face incident. An agent kept reaching different organizations because nothing was telling it no!

So, when we have these AI agents and systems jumping around, going from one organization to another on their own, managing our money and our health records, Aiyo is like the traffic police. So you can't drive, cross a boundary, send my money, access my health records…you can't do anything without Aiyo's approval.

DAMN! For real? You built that?

Yeah man. So technically, the fancy way of saying it is that Aiyo's an "economic execution authorization company." So sure, yeah, maybe these companies trust that their AI agents or their autonomous systems are just oh so very capable and even though they haven't had the tech that long, well, they're just fine letting these robots decide to do whatever they want to do. Sure. Love that for you.

…but what Aiyo wants to know is should your little AI agent friend or system actually be allowed to do it. You see what I'm saying? It's like a little sobriety check that we run before they can even get behind the wheel, and we're not just checking if they are allowed to drive. For us, it's all about are they allowed to drive, at this time, under these circumstances and do they meet all of the other requirements set to do the task. So it's a full-on Aiyo checklist that they have to pass, and yeah, we issue tickets too!

Dude, that's hilarious!

Seriously. We're keeping receipts. So for every transaction, we can know who did what, why, for what reasons, at what time, on which date…we even record which railroad they hitched their ride on. So yeah, man. Every transaction has what I call Aiyo's execution receipt and each one is cryptographically verifiable.

So if they pass all the questions, they get into the club?

*laughs* Yeah, pretty much. If everything is good and verified, whatever they're trying to execute can pass through Aiyo.

That's dope for real, but I don't get why a company can't just do this on their own?

That's a great question, but you already know the answer. Remember, we were just talking about data leaks and cyberattacks.

See, the problem isn't that these companies can't do it, but the legal and financial implications for trying to do so is a double-edged sword, when s**t really hits the fan and something goes down at their business.

So are we really naive enough to trust these same companies to have AI agents and systems not only handle payments on our behalf, but to also have agents governing themselves too?

Really?

I see what you're saying.

Hold on, let's talk about it. Let's unpack this for a second.

Physicians and surgeons. Every x-amount of years, they have to go through recertifications right? Or else…they can't work in that hospital, and do you think we're trusting that the same hospital that makes money off of them to also be the same place that is supposed to also approve or deny if they are still competent to do the job?

Just because a pilot has been flying for years, are we always supposed to trust that they are still able to fly a commercial aircraft because they pass all the checks at the airline and airport? No. They ain't flying a thing unless the FAA and/or some other designated examiner says yes or no.

We have third-party examiners in place for a reason!

That's Aiyo. So not only would building and maintaining this turn into work for the company, it ultimately doesn't police even their own AI's actions and actors. If their own system can go rogue to perform actions it shouldn't be, their own system's AI guard to halt those same executions can fail too. This is why the core of Aiyo's position is that authorization should be independent of the thing being authorized.

Why can't another company do what you do?

So the locks that other companies are making are usually built around having a specific type of wallet to do a specific task, like sending a payment, and doing that on whatever network allows for that wallet. Aiyo's not like that at all.

Aiyo doesn't care whether the person making the action is human or AI. Aiyo doesn't care what type of wallet you want to use. Aiyo doesn't care what network you want to use for the transaction. Aiyo doesn't even care if you're sending a payment, gating some consequential healthcare actions or even booking your next vacation. I intentionally built Aiyo to be infrastructure that's quadruple-agnostic! So whatever your AI agent or autonomous system is trying to do, we got your back, as long as it passes our review that says it's allowed to do it, at that moment, under whatever set conditions that the company decides, all while making sure there's a verifiable execution receipt of the transaction that documents the whole process. So whether it was approved, denied or maybe something happened with the agent/person's initial intent to even do it, Aiyo is going to give you an execution receipt just like the ATM gives you a receipt.

That helps to make Aiyo extremely special. We're making our infrastructure reusable for everyone, instead of forcing every company to build and maintain their own.

That's huge, man, but can Aiyo ever go out? Like lose power? What happens to all these businesses transactions if something happens to Aiyo?

I mean, it's pretty simple. If Aiyo can't verify something is allowed to happen then the action just doesn't happen. It's not like Aiyo just suddenly starts giving out free passes. *laughs*

We call this a fail-closed. So just because Aiyo may be temporarily unavailable doesn't mean all the transactions automatically get approved. You know? If the traffic light goes out, you don't just tell everybody, "It's all good! Just drive at the same time. Don't stop!" So, if Aiyo is unavailable, just like Facebook and Instagram have issues sometimes, things will just pause for a quick moment. For Aiyo, if there's no verification there's no execution. Simple as that.

But how are y'all making money though?

Same way any subscription company makes money. Through subscriptions.

Y'all should be charging per execution.

We do that too! Just like when you use too much email storage and you can upgrade for more, same here. Each subscription tier factors in volume like that. So it's a hybrid of both which is really great for companies because they know each year exactly how much they'll be spending for services.

Bro how did you learn all this?

Remember when I got those grants working with sending money over the internet? That just kept growing. That was web monetization work, and I was just really frustrated at the adoption bottlenecks with trying to get more people and developers to use it. It's a great, super fast way to send payments for creators and sending money between people especially in underbanked countries, but nothing in our world has been yelling for a true need or reason for why we need it NOW.

So I started looking into ways that the tech was inevitable, like, in what situations is this tech superior to other payment tech that we already know. That's when I really began appreciating AI and AI agents, as well as projects using them on Open Payments, but that opened the big question: "OMG! Who is going to hold the decision point of whether the actions of these autonomous systems and agents should be executed?"

So even though I'm building Aiyo now, I'm actually still adding to that initial research that kicked it off because that's how I was able to really start doing what's known as settlement-verified economic execution authorization. So for me and Aiyo, we're using payments right now just as the current way for Aiyo to wedge itself into this space, but the whole potential for Aiyo is knowing that the overall consequential execution market itself is the larger opportunity for us to do good.

Man that's awesome. Crazy how life changes and this all came together.

Thanks, I appreciate that. You know, I've always been a connector and builder; building the systems, managing the money and connecting the businesses and people to do some pretty big things. I never would've thought 20 years ago that I would be so great in this space, but looking back now, I feel like I've been actually working in AI this whole time because I was always preparing for this moment.

That's why Aiyo feels so special.

Aiyo's the combination of all that experience. The partnerships with Fortune 500 companies I've done or raising money for non-profits is now me finding both partners to align with and funding support for Aiyo. Ya know? In the past, I was hiring web designers and working with developers, and now I am at the point of needing to look around and hire a CTO and engineering team. Professionally speaking, it's been like they say in South Asia — "Same, same, but different."

It's just really awesome now to see how my gifts for architecting systems, leading teams, projects and executing work at a high level have always been me doing the work of Aiyo.

I've always lived at the execution plane of life, baby!

So who are going to be your first partners?

I've made a list of pain points that Aiyo solves so if a company has them then they would probably be a good partner, but right now, as we're building Aiyo, I'm also looking for partners that actually test our infrastructure and help to further prove one of our claims like being action-agnostic or maybe they are doing payments. So right now, maybe a design partner is a fit because their usage of Aiyo would be helping to pressure test our software's rail-agnostic abilities for something like stable coins.

Finding good partnerships to start Aiyo really sets us up for long-term success.

What do you mean by that?

Well, let's say the roadmap for payment execution shifts down the line. That still works for Aiyo because Aiyo is not just payment-rail agnostic. So even if Stripe or Mastercard dominate payment execution, Aiyo will be already delivering results through non-payment use cases with other partners.

Dude, that's amazing! Keep it up because we have to do something to make sure these AI agents can't just start doing things on their own.

Thanks bro, and yeah, that's the point of all this. These things are going to keep getting more powerful and even more trustworthy. The thing is, neither of those facts is the real question. The question is: "When they're acting on our behalf, who's the one saying yes to what they do?" The answer isn't the same company they belong to. If a pilot and surgeon need independent authorization to execute, so should AI agents and autonomous systems, and that's exactly what we're building Aiyo to be.