Trust in an Era of Programmable Money and Agentic AI

Transcript

A very good afternoon to everyone. Just a slight correction over there. I'm still a sitting member of parliament in my fourth term, but I'm here in my commercial capacity as co-president of Metacom, a MASlicensed major payment institution under our licensing capability. DPT is part of that. Now today I'm going to share some thoughts around this very important topic that I find uh very close to my heart and I would like to share my thoughts with all of you here. Um I believe that at this point in time we are at a very interesting juncture in the entire financial industry. So it is where we see the intersection of number one stable coins and digital assets becoming more mainstream especially within the

so-called financial institution or institutional finance and number two just as this is becoming more mainstream we are now going to welcome the next wave which is agentic finance um I would like to maybe break this down a little bit uh let me first talk about stable coins and digital assets now I think that in this case you know we've been here token SFFF various events it's been a few years running so stable coins overall together with digital assets has gone through several seasons first was the speculatory season where people were probably still thinking is this real I'm still very curious about it is it even going to happen then it became more of a

regulatory reckoning season whereby you see legislations and regulations around the world beginning to be more mature and also gaining clarity, pun intended there. Then the third one is the infrastructure season which is pretty much where we are right now and so the question has shifted from is it real is it going to happen to how can I access this or use this in a more compliant and regulated approach. So um in this case we can see this panning out here and if you look at uh what's happening it's not just regulations but in terms of international um you know G2G various types of infrastructure you are seeing

it. So FIS are also actively looking at adopting stable coins uh and also on the trade front because when we talk about payments we are a payments company so when we talk about payments it has to follow real trade activities as well. Let me use Azan Southeast Asia as an example. Now the Azan Deepa digital economy partnership agreement is finally seeing the light at the end of the tunnel. After years of of negotiation, senior officials in May earlier this year finally concluded on the terms and we are expecting it to be signed in November this year. Now why is this a useful point to think about? Because what this means is that there will be

some pipelines or you know the fundamental uh conditions being laid out uh for example interoperable crossber data uh digital payments integration e- invoicing so on and so forth. Now if you think about it from a different perspective, it is not just a trade agreement per se, but it is a mandate to build a shared infrastructure that can facilitate the use of the likes of stable coins and digital assets to facilitate real world activities and I think this is an important milestone that we should continue to embrace. Now the different countries within the Aza market have also been embarking on several projects whether it's a you know

with the government within their own ecosystem themselves or with private industry players we see more of such private public partnerships happening Singapore has got several projects to name a few but moving on from that is that if we look at stable coins market in general it has also been growing from strength to strength Not only has it been growing rapidly at double-digit pace, but also it is still very much on track to reach the 3 to4 trillion US by 2030. Now I believe that speakers before me would have said much about this but to us we see this happening on a daily basis. Last year stable coin payments transactions uh would have in total

reached 400 billion uh US dollars. And these are not just uh speculative type of transactions but really driven by real world trade uh crossber kind of payments. And for us at Metaccom uh we are very much a web 2.5 payments company from the infrastructure our core banking system to our ecosystem very much a hybrid natured uh company. What we have seen is that the entire of last year we are very privileged to share that we managed to hit a total of more than 10 billion US dollars involving stable coins and out of this entire you know 10 billion US pretty much most of it would

be because of crossborder payments and driven by real world uh trade between buyers and suppliers. So to us this is real and this is happening but to do so a lot of a few things will have to be settled and that means the infrastructure itself not just the tech part but also compliance because when we look at the traditional financial side the compliance that were designed were for the fiat type but on the on world but for the web three side we have a different nature. So in web two site uh for example if you use bank accounts identity is known but the transaction the history of transaction can be relatively opaque. Maybe you know your

direct counterparty or indirect one maybe one hop or if it's within the same banking ecosystem maybe you have more transparency whereas onchain that's quite different. Wallets could be anonymous to anonymous depending but the transaction history it's highly transparent. you can trace back infinite number of hops to understand where that token may have traveled over time and so you are able to establish patterns in that sense. So we will therefore need to build infrastructure to be able to reconcile all these idiosyncrasies within the web two world and the web 3 world. Now as we have managed to uh make breakthroughs in this aspect for in

terms of stable coins token the so-called web 2.5 world the next wave has broken and that's Agentic Finance. Now for agentic finance as we speak now today there are institutions out there already using agents to execute transactions whether it's with their customers whether it's within the internal organizations themselves or B2B so we have seen that happening now this is something that uh there's a lot of excitement we have been talking about this whether at token itself at site events outside but I feel that there is something that we will need to definitely address in order to see this uh adoption to scale and the very

important thing is trust. Trust in an era of programmable money and agentic finance. So why are we talking about this now? Because now we see the convergence of programmable money and agentic AI coming together. It's almost like a perfect match in heaven. programmable money such as stable coins, tokens, they can settle instantly. Uh it can you know be 24/7. We can avoid you know having the corresponding bank friction and so this is something that is and it can execute programmable conditions as well. So these are the unique characteristics of this. Then we

have agentic AI, agentic finance where you have agents that are autonomous that can work 24/7 as well and that can therefore use programmable money which has finality has speed at atomic level. Together you have a genuinely powerful finance capability. You have finance that can think, act and settle in one single continuous motion. So that's great in most instances because it means that higher efficiency hopefully lower cost because you optimize everything.

But what this also means is that decisions made can be irreversible. So if you have a good agent, that's all great. But if you happen to encounter a bad agent making that decision for you, then we'll be in trouble. And so this is where the question of governance comes in. How do we build trust into this system? Now, if we think about this um and if I think many of you would have been uh watching the news following some of these uh international developments, uh AI safety has been coming to the forefront of many conversations around the world. Even for Entropic in its most recent IPO filing, hundreds of pages,

there's a significant portion of this dedicated to warning people about the use of AI even as it talks about the opportunities of AI. So for example, one of this would be that uh agents could get so intelligent that it can be cunning. Cunning in what sense? that it will know how to mask its track to behave differently if it knows it's being monitored. Other international bodies like the FSB has also said earlier this year that the governance structure will also have to shift over time because of the nature of agents working 24/7 at such speed. Um it's almost impossible for humans to check

every decision point or every step of the way. So the whole construct will have to change. All policies that were designed for the humans to use systems and machines to execute something will not quite apply in this context. Now if we also look at some other I mean this is not this is this is not something that is just cooked up by these two entities that I've just cited. If you look at uh EOS it has also said something similar. There are other jurisdictions such as EU around the world also saying something similar. Now for in most recent time I thought something that is very interesting to to to think about is that um even at the UN level uh this is also very much a

conversation Singapore foreign minister Dr. Vivven Balak Krishna at last week's UN General Assembly spoke of and called on uh having a UN convention uh on AI safety especially for agents and that an international neutral body be set up to promote or set common technical standards promote best practices and also to see how we can uh in a way uh promote uh robust testing and assurance to this so that there is something that is neutral uh a body that is neutral to be able to certify this. I think these are all very significant and important milestones and we should sit up paying attention to this. Now for us at

Metaccom, we believe this as well because as a licensed payment institution, we believe that not only do we want to deliver payment services that build resilience into finance that offer our clients options, but it is also important that we embrace frontier technology and deploy them in a responsible manner. So while we have now moved forward to into Agentic Finance itself, we feel that it is important that we make sure that the governance of a of the agents that we use are in place so that we can build confidence and build trust amongst our partners and most definitely instill that confidence in our clients. So what we have done is

that in April 2026 this year by the site of money 2020 uh 2020 we launched our stable X know your agent framework the KYA framework. So for this framework we took reference from several guidelines. We took reference from Singapore statutory board IMDA's model governance framework for agentic AI first launched in January later updated in May. We've also taken reference from MAS SAFR framework and as we speak now we continue to ensure alignment with several international standards and frameworks as well. Uh so this is KYA. Now for KY we have

organized this in according to four pill uh four principles and four pillars. Allow me to take this chance to share a little bit more about our framework with you. four principles. First principle assess and bound and that is that we must know what use case or use cases are we using the agents for. The agent can do many things but we need to define what uses this agent is meant for. Why is this important? Because once we know the use case, we'll be able to understand the risks that may come along with the activities that the agent will have to engage in. and therefore with that we'll be able to put in place the corresponding mitigation measures.

Number two, human accountability. Now technology should be an enabler to help humans do our work to live our lives better not to replace us. And therefore human accountability remains an important part of this. We we shouldn't be voding ourselves out of the equation basically. And what this means is that every agent should have at least one human if not a team uh bound to it to be responsible and accountable to that. But it also means that at critical decision junctures and explicit human approval should also be designed in to the process. Now the third principle technical controls. This is where we

introduce something deterministic into what would have been highly probabilistic. So technical controls is important. Give a very simple example. For example, your agent could autonomously execute your instruction. So I may instruct the agent to say that now I want to buy a pair of blue shoes. But clearly I wouldn't want the agent to buy something that is $10,000. So I may set a threshold to say that uh below $100, please go ahead. Above that, please check in with me. The fourth principle is user accountability. Now this is important because the user himself or herself should know that he or she is still accountable for the use of the agent but of course the rest of

us will also have to do our part to support the user. And then for the four pillars, agent identity. This is important and the most important part of our framework because the agent is autonomous and what we have built in is that the agent will have to have a unique identifier and be stored in a database that is verifiable. And then we also have the scoping permission control as well as behavior monitoring ensuring that every activity is being locked and in an immutable manner. And the last pillar is about how do we govern agent to agent interaction because we don't want that any rogue agent to enter the

system and then retrieve information or do something that we did not authorize it to do. So if you think about it for the agent identity after verifying it after clearing the assessment we issue a certificate to certify that this agent has passed the test and then with that assuming it's like a trust mark when the agent starts interacting with other agents they will then have to exchange credentials to make sure that they are mutually recognized before the next step of action can take place. So for us this is something that we are very proud to have put in place and we continue to want to work with industry players like-minded partners to come and jointly continue to build on this and to you

know uh as an industry uh I hope that we can build on this to create a common set of standards. Now I'm mindful of the time. I know my time is up here but uh get a little bit got a little bit carried away in my sharing. But for us what we have done is that we have now not only upgraded our KYA but translated it into real live production and running. And so this is something that we have uh in place. And um for this if I could therefore to save time here I would like to end with a call and a video. Call is to again to reinforce my call to everyone that we need to

continue to work together because it is when we come together and not behind wall gardens that we can ensure that there's interoperability. With interoperability, adoption can scale. And so if we want to continue to go forth with agentic finance, we need to be able to mutually recognize each other's standards even if we do not adopt the exact same solution. So this is important and the thought leadership on this front will also continue to have to progress and so this is my call to everyone to join us in this effort. And finally, a video to showcase to you what we have done in KYA and what is already

possible live in production. With that, I want to thank you for your time and uh may I bring up the video? Thank you. >> AI agents are starting to act on their own, checking accounts, moving money, making decisions. But when an agent acts, how can a regulated institution trust it? And who is accountable for what it does? These are the questions that stable X know your agent [music] launched in April 2026 is designed to answer. Stable X Aentic [music] network connects AI agents to financial institutions within a regulated safeguarded framework. Built with interoperability in mind, it lets agents run with our SDK or on agent XC cloud. Every agent runs in its own

isolated environment with its own identity on the stable X agentic network. At the heart of KYA is the KYA certificate, the agents passport. It's a trusted credential that institutions can verify in real time, aligned with IMDA's model AI governance framework for Agentic AI and with MAS's safeguards for Agentic Finance at runtime, which sets tighter guard rails for AI agents operating within financial systems. A KYA certificate isn't just a badge. Each agent is bound to a customer the institution has already verified through KYC. It carries clear spending limits, a named responsible person, and a backup contact, and it's suspended

automatically if left dormant. The user gives informed consent before anything goes live. On the other side, institutions see every agent registered with them, monitor them against KYA metrics, and decide which agents receive a KYA certificate and which don't. Once issued, the KYA certificate becomes active and the agent can access Metac's authorized financial services within its defined permissions from checking balances to making payments within its spending limit. Here it verifies its KYA certificate, then retrieves the balance through Metacom's authorized tools.

That's KYA on stable X agentic network. An identifiable agent, defined authority, human accountability, and institutional control. AI agents that institutions can trust. And there's more to come. We will keep building towards interoperable, trusted, agentic finance. We invite like-minded stakeholders to join us, collaborate, share [music] best practices, and help shape what comes next in Agentic Payments. Metacom, bringing you easier payments, resilient by design.