Full transcript: Raghu Bala on London Real
0:00 Brian Rose: This is London Real. I am Brian Rose. My guest today is Raghu Bala, the technology entrepreneur, educator and founder of Synergetics, a company which is powering the tokenised AI agent economy with the world’s first patented A2A communication protocol. You’ve held senior roles at Yahoo, InfoSpace and PwC, had four successful startup exits, and hold an MBA from the Wharton School and a master’s degree in computer science from RPI.
You’re also the head managing instructor for 2U’s MIT Sloan School AI and blockchain programmes and an adjunct professor at VIT in India. At Synergetics, your mission is to empower the agentic AI ecosystem, a world where autonomous AI agents, digital twins and devices can interact, transact and learn. With the upcoming launch of the SGTX token, Synergetics is helping build the financial and technical backbone of the agent economy. Raghu, welcome to London Real.
Raghu Bala: Thanks for having me, Brian.
5:01 Brian Rose: It’s great having you here. I want to start by saying that Synergetics is one of our portfolio companies here at London Real Ventures, the world’s first media-powered investment firm, where our mission is to transform the world by investing in cutting-edge Web3 and AI companies while giving our partners exposure across the London Real media platforms. We heighten awareness, create customers and build communities by leveraging our global audience and network. Raghu, we are super excited to be part of this journey with you. So let’s jump in. What is Synergetics?
Raghu Bala: Synergetics is powering the agent economy. All of us have seen other economies develop. For example, if you look at YouTube, it’s a meeting place for creators, advertisers and consumers to come in and they all transact. Consumers consume content, creators create that content, and advertisers help monetise that content by sharing the proceeds with the creators. We are seeing the same movement happen again, but this time around for agents.
Currently, everyone has probably used ChatGPT. That’s like the first real consumer version of AI. AI has been around for a long time; I’ve used it even during my Yahoo days. But for the everyday person, AI was not within reach until ChatGPT came around. ChatGPT is a simple application that’s built on top of a model.
Think of AI as electricity. What came after electricity was light bulbs, electric trains, television, motion pictures, all of those powered by electricity. The way I look at it, foundation models are like electricity. They give us the building blocks. Now what do we do with it? And agents are the first manifestation of that electricity in a kinetic form where it can start to do things.
Agents currently are being used by enterprises within the organisation. But that’s a short-term trend; the beginning is the foundations. What’s going to happen next is going to revolutionise how everyone is going to see agents.
7:29 On the consumer side, you and me are going to create our own digital twins. Our twins will be replicas of us. They’ll look like us. They’ll talk like us. They’ll speak in many more languages than you and I even know. They can be our alter ego online and be able to conduct transactions on our behalf.
So Brian, you might be running or practising for your next marathon, but your twin could be shopping for you while you are busy doing other things. You have empowered your twin to do various things and it’ll complete those tasks, so you can be so much more productive. The same thing with companies. Companies are going to have twins, or their agents, able to transact with other agents outside the enterprise, with their suppliers, their partners and so on.
For that to happen, you need the agent economy in place. And that agent economy needs many building blocks. That plumbing, those building blocks, are what we have built in AgentWorks, our flagship platform.
8:43 Brian Rose: And is that based on the patent that you have on this A2A technology? It was one of the reasons we were really drawn to you, because there’s a lot at stake here. If you get this wrong, you could ruin companies. You could lose billions, I would think. So this exact protocol is important. Maybe you could tell me a little bit about the patent and how your tech is different.
Raghu Bala: Sure. One Sunday morning, I think in 2024, I woke up. I don’t know, maybe I had a good night’s sleep. I woke up with this epiphany. I said, “Oh, I can see the future.” I know exactly when this happened, because for the next four hours I was just feverishly typing all of my ideas, which became the patent. It’s a set of building blocks that you need in order to make the agent economy happen.
9:38 The first building block is that agents need to have identity. Without identity nothing else matters. But that identity needs to be registered somewhere, and that’s where the registry comes in. The registry is like the old-fashioned yellow pages where you look up people’s names. Here you look up the ID of the agent.
Why do you need that? You need that for a process called discoverability. It’s a very crucial part, which is why we are partnered with MIT Media Lab on these registries. It’s a funny quiz I give some of my co-workers. If you have one person, you need no connections. If you have two, you have one connection. If you have three people, you have three connections: A connects to B, A connects to C, B connects to C. If you have four people, it’s six connections.
It’s called n times n minus one over two. It’s an n-squared problem in computer science, which means it’s got a complexity of n squared. When you have a registry, that complexity gets down to one. If I want to connect with you, I don’t need to know your address a priori. I just have to look it up in the registry, get the address of Brian’s twin and then connect with Brian’s twin. I know it gets a bit technical, but it’s very important because it reduces complexity.
The idea is that all the agents in the world will ultimately register themselves with a registry. We have our registry. Other companies like Google and Salesforce and ServiceNow are all creating registries. So the folks at MIT and us got together, and one of the things that we decided to co-develop was a registry of registries, which is like a yellow pages of yellow pages where all of the world’s agents can be traced. If anyone wants to talk to anyone, they can look it up, and that becomes an index very similar to how Google is an index of web pages. This is an index of all the agents in the world. So that’s a critical building block.
12:00 So you have ID, you have registry. The next thing: just like you and I carry our driver’s licence in our wallet, the agent has a wallet. We normally carry credit cards or currency in our wallet so that we can go out and buy a cup of coffee. In the same way, the agents have got a wallet which has cryptocurrency or stablecoins. That’s the third building block.
The agent also has a name. We have the lock on .twin, so you can have Brian.twin. That TLD belongs to us; we have partnered with Unstoppable Domains on that. The next step is that this agent will have to communicate with websites or with other agents, and this is where A2A comes in. Websites you can connect to via the protocol that Anthropic produced, which is becoming the standard for the industry, called MCP, Model Context Protocol. With other agents you need to use an agent-to-agent protocol.
This is where we had our patent approved in February of this year. It’s called AgentTalk, and it describes how two agents will communicate with one another. The first level is authentication and authorisation, to make sure that you are who you really claim to be. Whenever we add agents to a registry, we go through a process called KYA, know your agent, similar to the KYC and KYB that are already there in the banking world.
The second level of the protocol is vocabulary. If you go to a maths class, the conversation is about maths. But if you go to a soccer practice, the conversation is most likely about soccer, nothing to do with maths. There are vocabularies in every conversation that happens. So we have created an extensible set of vocabularies, but we have given it away as open source, in the sense that any company can extend vocabularies. You can build whatever vocabulary you want.
A canonical example that we have is that these agents can not only manifest themselves in the virtual world, but they can also manifest themselves in the physical world. A car can be an agent and a parking meter can be an agent. The same car can be an agent with a toll booth or a fast-food drive-through. But those three vocabularies are different, because the parking meter only understands parking, and the fast-food restaurant will only understand, let’s say, Wendy’s menu and that vocabulary. These vocabularies will become numerous over time. That’s the crux of our protocol.
I’ll segue into something which is kind of interesting. I saw Google’s VP give a talk at the MIT Media Lab event in July, and we approached him. He said, “You guys introduced A2A before we came to market, but we want to grow in the spirit of collaboration,” because there’s no use for multiple standards to be in an industry. It’s better to coalesce towards something that everyone will use. It’ll be beneficial for the whole ecosystem. So we have come together to try to extend A2A. There are benefits of A2A, and we have certain unique attributes and benefits. We are now working closely with Google and other players in the market on agent payments. That’s a huge area, but I’ll stop there.
16:15 Brian Rose: This is a huge piece. I’m good friends with Yat Siu of Animoca Brands, and I’m good friends with Dan Tapiero, who runs a whole suite of funds called 50T and invests billions of dollars. Both of them have been telling me for years that they were really bullish on blockchain because it was the only mechanism where AI and AI agents could compensate each other with these microtransactions and fast transactions. They’ve been talking about this for years before I really saw it coming into practice.
I remember I asked you a question on our Zoom call, and I ask this question a lot to our potential portfolio companies. I think I said, “Raghu, why screw this up by putting it on the blockchain?” And you came back with a really powerful answer of why you needed payment rails on the blockchain. Could you tell us why the blockchain is not only important, but maybe the only way you can have these agents communicate with each other?
Raghu Bala: There are some new protocols that have come about even since we spoke. It’s kind of incredible. But first of all, let me cover blockchain. If you look at the payment rails of the traditional banking system, the settlement time is two days. You’re not going to have a car talking to a parking meter, which needs to be paid now, and tell the parking meter, “I’ll make sure the money gets to you in two days.” That’s not going to fly. You need something which is immediate.
One of the things that afflicted the blockchain space was gas fees. Those in the Web3 world know about these fees. They were somewhat expensive, exorbitant at times. Now there’s a new protocol called x402, which is on-chain, but the transaction fee is something like 0.001 cents. This is becoming very popular now, and I believe that it will become a primary way in which agents pay other agents.
Brian Rose: Okay.
Raghu Bala: Because with these microtransactions we’re talking about a few cents. You cannot have a gas fee of ten bucks for something which you’re paying a few cents for. It makes no sense. It has to be almost negligible, and that’s what x402 does.
Brian Rose: That’s an Ethereum protocol?
Raghu Bala: I believe so. Coinbase is behind it. But the fact remains that blockchain is the only way to establish these payments, because it’s the only universal way in which agents anywhere in the world can pay. Otherwise, if you look at the banking system, when you go to different countries they all have correspondent banks and get to SWIFT in some way, shape or form, but it’s many layers and a lot of charges. If you send funds over certain cryptocurrencies, it’s a few cents, and if you send it over your bank in the US, it’s $45 per transaction, regardless of what I’m sending. It’s a non-starter.
So blockchain is the only way, and within blockchain we have to go down to the least expensive way of doing it. These microtransactions are going to be numerous. In fact, they’re going to outstrip the number of transactions that we see with humans involved in payments.
Brian Rose: Right. And I think you also said that on top of that you’re looking at a 3% transaction charge for everything on credit cards and debit cards. So not only is it slow, it’s just prohibitive.
Raghu Bala: Yes. There are three layers there. There are the banks, then there is a gateway fee, and then you have the providers, which could be your PayPal and other players. All of them take a cut. When you remove the middleman, what is 3% actually goes down to 0.54%.
Brian Rose: Okay.
Raghu Bala: And now I’m seeing that with x402 it’s even lower than that, way lower. The 0.54% is when you use stablecoins, but I think x402 even beats stablecoins in terms of payment. Mathematically, it’s the only way to do it.
21:43 Brian Rose: You mentioned blockchain. Are you working with any of the blockchains, and if so, is it Ethereum? How does that look right now?
Raghu Bala: Right now we have the world’s first agent wallet, which can be used by the owner of the wallet and his proxy, which is his twin. The same thing applies for B2B as well. This wallet is multi-chain. We have Bitcoin, all the Ethereum variants and all the Solana variants, and now we’re adding Avalanche, because Nasdaq is apparently going to tokenise stocks on Avalanche. There’s a mobile app on iOS and Android that anyone can download, and the agent version of it is an API version running behind the scenes.
The registry that we have, we started off with Polygon, but we’re moving it to Base, because Base is beginning to be a major player in that space and we want to be as mainstream as possible. We don’t want to give anyone any reason not to adopt these technologies, because it’s too important a wave that’s coming.
Brian Rose: I think Base is a really smart move. We speak to a lot of startups in Web3 and we fund a lot of startups, and right now when I talk about where they’re going to launch their token, it’s almost always Base number one and perhaps Solana number two. What Base gives you is, first of all, an amazing layer 2 on Ethereum where you can transact fast and cheaply, but it also potentially gives you access to every Coinbase user and maybe every Coinbase institutional user. They’re doing custody for a lot of big entities, including, I think, the US government’s crypto, and they’re pushing it with their own social network. So I think that’s really smart.
23:43 When it comes to agents, we’ve all heard these great dreams that an agent is going to do all these things for you, but we haven’t always seen that in practice. I know you talk to companies like Google and Deloitte and these other big companies that, like you said, are using agents internally, but that’s only going to get you so much. You’re going to try to help them send those agents out into the real world, where things get dicey and dangerous. Can you talk about some of the challenges that the industry is facing with that final step of putting the agents out in the wild in these B2B plays? Because that’s where you unlock a lot of value.
One of the reasons we decided to invest in Synergetics is that, unlike most AI plays in Web3, you’re not only selling services to the Web3 players. A big mistake people make in crypto is they build crypto tech for the crypto bros, realise there’s no audience, and then the company goes nowhere. When you said, “We’re talking to Google, we’re talking to these huge institutions,” that’s what got our attention.
Raghu Bala: I concur with that. The issue is this. When I go to the Web2 conferences and the Web3 conferences, there are almost two religions at play: the AI guys, and then at the Web3 conferences the crypto guys. Both are talking about decentralised AI, but they’re seeing different things. So we set out to ask what are the things that will easily see mass adoption.
Decentralised AI is made up of three types: decentralised compute, decentralised data and decentralised identity. When we talk to enterprises, I know for a fact that they don’t want decentralised compute and they don’t want decentralised data. They’re very protective of their data and they want to run it on machines that they control. So decentralised compute and decentralised data don’t work for them, and we said we won’t go that route. But they’re okay with decentralised ID.
If you take any big company and look at their workforce, I’d say 90 to 95% of the people in the company work internally on things like finance or marketing. But about 5%, the C-level people, the salespeople and some marketing people, have outreach with external folks. The same thing is going to happen for agents. You’re going to have this bifurcation: agents that work inside the company and will not interact with anything outside, and a few agents which will have contact points with external people.
Take a finance department at a public company, for example. Usually the auditor is an external company like PricewaterhouseCoopers or Deloitte or KPMG. There’ll be some people in the finance department who interact with the auditors outside. Imagine an agent on your enterprise side and an agent on the other side. So there is an epiphany happening where companies are starting to think, “We are using agents internally, but soon they’ll have to interface with outside agents.”
I had an opportunity to talk to the chief AI officer of Fidelity recently at a conference in an open setting, and I asked him this very question: are you starting to interact with agents outside your enterprise? He said there’s definitely a movement towards that. First of all, they’re doing it within their various divisions. They almost run these divisions as independent entities, so that’s agent-to-agent communication among the divisions, and soon they’ll have communications with agents outside their company.
27:58 The example I always give is from a US context. I can use my driver’s licence, go to the airport and fly anywhere within the US. But if I want to go outside the US, for example I’m here in the UK today, I have to carry my passport. In the same way, think of the internal ID that is used within the enterprise as your driver’s licence. Agents inside the enterprise trust each other and can communicate with one another. But when you want to talk to agents outside, now you have a trust problem. You need a passport.
So what is that passport? We started to create this registry where you register these agents, and then MIT’s initiative, NANDA, came along and built this registry of registries. Now, speaking of Consensys, there’s a new ERC standard called 8004. It’s called Trustless Agents. It’s exactly what we built, but they have standardised it. So now we are speaking to them about participating in 8004 and at least becoming a reference implementation of it, because we’ve already built it.
Trustless means that no agent needs to trust another. It can look it up in a registry, and by the fact that you are in that registry, any two agents can trust one another. What we built is now becoming standardised, and that’s a very healthy thing for the market. It also gives you this property of discoverability, so anyone can look up anyone and go and contact them. If you don’t have that, you have the peering problem I mentioned earlier. Once you have billions of agents, that’s an intractable problem. So you have to simplify it, and 8004 is the right step at the right time. We are glad to be working with the likes of Consensys and Ethereum on that.
30:19 Brian Rose: So you’re not out there trying to defend your patent and get people to stop using it. You’ve created this agent-to-agent protocol, and when other people want to adopt it, you say, “Let’s welcome this. We want to be the standard.” You just stay ahead of the innovation curve, and you’ll grow and reap the rewards just by being on the front lines there.
Raghu Bala: Exactly. We’re talking about some really well-established companies. We are still a startup and we need to know our place in the world. Obviously we have lofty goals. We want to be a unicorn, and every startup should aim for the stars. But we understand where these other players are and how much power they wield, and we have to work collaboratively with them. That’s the best way to do it. Otherwise you create these islands of technology that don’t interoperate, and there’s a problem.
In fact, there are a lot of agents which have already been developed with tools like CrewAI, LangChain, Gemini and so on, but the number one problem that they have is they don’t talk to one another. So we have given away to the ecosystem something called Agent Wizard, which gives these agents the same properties that our agents natively have: ID, name, wallet, the ability to connect and so on. We’re doing this because we believe that if people adopt these capabilities, over time we’ll benefit, either directly or indirectly. We cannot tell everyone to switch over. We’d rather be like the O blood group. You can accept anything. You don’t need to force people to adopt your way of doing things.
Brian Rose: This is the perspective of a multiple-time founder talking. There’s that old adage that first-time founders focus on product and second-time founders focus on distribution or collaboration. I know this is five and six and seven for you when it comes to companies you’ve started. The young kid starting off says, “It’s got to be my way, and everybody else is copying me.” You’re thinking the opposite: “We want to give these protocols away. We want to be the standard. We want to move things forward.” It’s a whole different mentality, isn’t it?
Raghu Bala: Right. The way to really build market share is to be friends and partners with everyone, and that’s our approach. You want to be in the middle of the conversation. You cannot make enemies. You have to collaborate, because the space is so broad and it’s growing. By the end of this interview there’ll be two new things that have come up that I have to go and read up on. That’s how fast it’s moving. So our approach is to be woven into the fabric of everything that’s going on, and I think that’s the right approach.
33:53 Brian Rose: What are the next big milestones for you? Is it about landing those big clients like Google and Deloitte and getting into their ecosystem and providing them the service, or is there something else that you look at as a major metric to get you to the next level?
Raghu Bala: These collaborations with Deloitte, HP, Google and others are a means to an end. The end customers would be the large enterprises, and they are the ones who need to adopt this technology. That’s on the B2B side. On the B2C side, it’s equally exciting. I’ll just throw this out there to you. There are seven or eight billion people on Earth today. Do you know how many phone calls we make to each other per year?
Brian Rose: How many?
Raghu Bala: Five trillion.
Brian Rose: That’s crazy. Five trillion calls between seven or eight billion people.
Raghu Bala: Exactly. We used that metric because when we had to come up with the market size of what we’re talking about, we went back to MIT Media Lab and asked how many agents they think will appear in the market over the next five years. Their number is one trillion. Seven billion people, and they think one trillion agents. Then we said, let’s be a bit more conservative and say 100 billion. It’s still a big number. It’s almost ten times the human population.
35:43 Another interesting metric is how many devices you have. I’m talking about the developed world; let’s say a billion of the people on Earth are in the developed world. How many devices do you have?
Brian Rose: This would be a desktop, a laptop, a phone and other wireless devices. I’ve got an Oura Ring. It’s a lot. I have a lot of phones, man.
Raghu Bala: I’m in tech, but I’m probably ten. Ten is a good number. If you just take that, and I’m only talking about one billion people and ignoring the other six for now, that’s already ten billion devices. One of the conversations we’re having is with Samsung Research. We are working with their team.
Brian Rose: Samsung.
Raghu Bala: Yes. All the appliance companies, whether it’s LG or Samsung, have publicly said that all of the devices that right now are called smart devices will be agents. All your cars are going to be agents. I was the CTO for Automotive.com. There’s a rumour, and I think it’s true, that Tesla cars have had wallets built in since 2016. I cannot confirm or deny that, but I do know for a fact that Daimler introduced a wallet that can be plugged into cars in 2018. So these cars are going around with these wallets, but no one to talk to.
Now that the Wendy’s drive-through in America is actually AI-powered, it can be an agent. So theoretically it’s not a very distant hop to say that a self-driving car can go through the drive-through, order Brian a burger, fries and a Coke, and come back while you’re shooting something in the studio. You don’t need to call DoorDash or whatever service that is in the UK. That’s the level of automation we are talking about.
So the ten billion agents for one billion people times ten is one side of it. But every car, every parking meter, every toll booth, everything around us is going to become a smart device. In fact, we have already proven it. We have a Jetson Nano, which is a small piece of hardware that Nvidia puts out, and we run nanoGPT on it. This can be added to any other device.
38:29 We are in the middle of setting up a proof of concept for Carrier Corporation, which is one of the big companies in cold-chain supply chain, refrigeration and air conditioning for buildings, an HVAC company. We’re still in early preliminary talks, but one of the things that they are interested in is putting agents within every truck to ensure that the food or the pharmaceuticals don’t get spoiled.
That might seem like a very simple problem, but it’s not, because every item that you carry in such a truck has a different temperature at which it will remain fresh or it will spoil. Too cold and it gets frozen and brittle and it might spoil. Too warm, it might also be gone. So you have to be very careful. They don’t only want smart devices, but smart devices that can think quickly and without human intervention.
They do business throughout the world: China, Mexico, the UK, Europe, the US. These vehicles might be anywhere, and they want to make sure that there’s intelligence in those remote places. This idea is not only for one company. Almost every company out there in logistics would have to start thinking in this way. So I’m saying this 100 billion agents is not a big number if you start thinking a bit broadly.
41:14 Brian Rose: Right, and the scale just keeps going and going. I saw recently that rugby players here in Britain now have smart mouthpieces, and a light will go on if the player has received a concussion. Apparently they used to have them where you could test afterwards, but now it’ll start blinking so the referee and everybody can see it, and they have to pull the player out. It’s yet another introduction of a device that probably needs to be an agent.
Raghu Bala: Yeah, and there’s going to be so many. Every human is probably going to have hundreds. It’s funny you brought up sport, because as I told you, I’m a British soccer fan. Sometimes players, after they score a goal, take off their shirt, and I always saw them wearing a small vest-type thing. Every player wears this thing. I discovered later on that it has got a tracking device at the back which shows their movement, how much ground they covered in the game, how many passes they made.
Brian Rose: And vitals as well.
Raghu Bala: Yeah, it does everything. Almost every professional team now has a sports science department that analyses player movement: how much ground they’re covering, how many passes, how many headers. So it’s interesting that you mentioned that.
Brian Rose: And in the NFL they’re all miked up now, and with the various subscriptions you can go and listen to what each player is saying. So yeah, the future is going to be filled with agents.
43:20 I want to ask you a question. You’re a head learning instructor at the MIT Sloan School, and you spend a lot of time at the Institute. I spent four years of my life there, from 1989 to 1993, and I knew at the time how special that place was, but as I get older I value that experience even more. There’s something about MIT that is different from any other place in the world.
I still remember, even when I was a punk kid getting a little cynical about school in my junior and senior year, walking into Lobby 7 on 77 Mass Ave, looking down the Infinite Corridor, walking by all these Nobel Prize winners, and going to study at three in the morning in one of those rooms. I just knew I was in a special place. You continue to go back there, and I know you’ve met the Google people there and many others. What’s it like being associated with that institution and being able to meet with all the people that come there? How are your interactions different at MIT than they would be anywhere else?
Raghu Bala: First of all, I don’t go there with a commercial mindset. I go there with an academic, learning and research mindset, and to me it’s very nourishing. Every time I visit these events or conferences or get-togethers, it’s like what Steve Jobs used to say: I get a lot of dots and a lot of ideas, but they don’t connect at that time. Then maybe days, weeks, months or even years later, they start to connect by themselves, or maybe I’m sitting next to a person on an aeroplane, I have a conversation, and I put that dot together.
When you get these really bright minds together in a room, it’s just incredible what happens, because that starts to sprout or germinate into various branches of ideas that you then connect later on. You also get to know what other people are thinking about, where they are going and what they are doing. In fact, three research papers on agentic AI were recently published that I had the privilege and honour of co-authoring with some of the researchers at MIT. There were also external participants from, I believe, Salesforce, Cisco, Carnegie Mellon and so on.
It’s really interesting to collaborate. We have these working groups where we throw around ideas about how this space will start to evolve. If you don’t open yourself up to these inputs, you’re actually losing out. These events are surprisingly inexpensive or free, and anyone can watch them online. So I’d encourage anyone, whenever these things come up, to follow NANDA, which is the Internet of AI Agents. All of us are going to experience bits and pieces of this in the coming years, so I would encourage everyone to at least follow at a cursory level what’s going on.
Brian Rose: It must be nice to take off your commercial hat for a little while, go into a research and ideas exploration hat, and then go back to implement it.
Raghu Bala: Yes, and implement it. One of the things that I take pride in is not just talking the talk or even writing the papers, but actually trying to find real-world use cases immediately. You can write anything you want in academia, but only when you implement it will you come across the economic barriers. There are a lot of things in science that you can do on paper, but in reality they’re too expensive. So implementation is a good teacher. It tells you this is good theoretically but practically it’s not going to fly. When you start to implement, you might tweak it and say version one might only go this far. It gives a different way to bring it to market, and a different skill set is needed.
Brian Rose: Very important. And then you’ve got to sell that thing and see if anybody actually uses it, and then see if it can make money.
Raghu Bala: That’s a whole other world. Absolutely.
Brian Rose: One of the reasons I was drawn to engineering at MIT is that I always liked the practical side. I had friends who were physics majors and other majors, but when you’re an engineer, there was a famous course at MIT called 2.70 where they give you a box full of wheels, gears and motors, and you have to build something that’s going to fight another student’s project. It’s all public, and they’ve been streaming it live since I was there. You literally battle these robots out. They’re probably implementing AI technology now. It was a real difference: you learn about how these things work, but when you have to put it into practice, you realise that a lot of that goes out the window.
Raghu Bala: That has now evolved into swarm bots, intelligent swarm bots. These swarm bots collaborate with one another to solve a problem.
Brian Rose: That’s the next level.
49:39 Tell me what your next 12 months look like. What is your single biggest challenge now running Synergetics, and what are the big milestones? You’re talking to a lot of the big players and they’re recognising that they need this agent-to-agent protocol, but what are the next big things you have to solve?
Raghu Bala: First, what we’ve done is grow from the inside out. We’re working with the large companies to implement agents at scale. In the first couple of years, companies started to experiment with agents, but they have all been singular agents, operating on their own, and that’s quite trivial actually. Now a lot of them are gravitating towards multi-agent systems, that is, agents collaborating with one another or with humans. That’s more the type of thing that happens in companies daily. The next step is education. We have to educate these enterprises to move towards agents communicating across enterprise boundaries, and the 8004 standard and other industry infrastructure being in place enables that transition to happen.
As far as immediate priorities, this quarter, Q4, we are actively in discussions with a lot of companies, because I was C-level at many companies prior to this, and this is the time when they set up the budgets for the next year. Companies set up budgets in two ways: earmarked line-item budgets and discretionary budgets. We want to get into those earmarked budgets, because that will almost guarantee you revenues for next year, and in business revenues are the lifeblood. That’s what we are really focused on in the very immediate short term.
The second thing is that we are planning to launch a token. We are in discussions and moving that along, probably with an eye towards seeing something next year, and that’s got its own trajectory. Then there is fundraising on the equity side to help strengthen the company’s balance sheet and funding, because stability is important in the sales cycle as well.
52:31 Apart from that, we are keeping an eye on a lot of technology changes and our roadmap items. One of the areas that we are really focused on is building out a marketplace. The marketplace is where the four big constituencies in the agent economy come together: enterprises, creators, community and consumers. All of them coalesce in that marketplace, similar to the Amazon marketplace. Enterprises come there either to rent agents that are pre-built or to hire community members for testing agents and other AI-related tasks. Creators are there creating models, selling data, selling pre-created agents or selling skills.
For example, a consumer might say, “I need a skill for my digital twin to be able to book travel,” but you are not a developer. You just want to buy that skill, similar to renting a movie on Amazon Prime. So you rent a skill. That skill might have been developed by someone who knows the travel industry well and has built all the APIs into the major airlines and travel sites. They put the skill out there, you rent it for maybe a couple of bucks a month, and now you can have your agent go and book travel.
The ecosystem comprises these four entities, and we are actively launching the marketplace that’s going to be the focal point of everyone coming together to transact, and that’s where the use of the token comes into play. That’s a primary thing that we’re focused on. Hopefully, I’d say end of this quarter or early next quarter, we’ll launch that.
54:36 Brian Rose: Why is a token necessary? And once you generate that token, what will be the utility of that token?
Raghu Bala: It’s a utility token, with many types of use cases. First of all, especially for the creators, they can earn. We want to create an ecosystem where people not only spend but also earn. A lot of people ask me, isn’t AI going to replace jobs? But AI is going to create jobs, and where is it going to create jobs? In marketplaces. Before YouTube became a big thing, let’s say in the early 2000s, there was no such job as YouTube content creator.
Brian Rose: Yeah.
Raghu Bala: But there are a lot of people who make a living on YouTube; their income comes from creating content. The same for Instagram, Reels, TikTok and so on. This agent economy is going to create a lot of new jobs, and for that economy to function in a rational manner, we need a token. That token will incentivise all of the players to participate. And I’ve not even covered the makers. People in the maker ecosystem, like the bots you mentioned at MIT, can create all kinds of devices, medical devices or drones, and all of those things will become agents. That community can participate in this too, so there are a lot of use cases that will derive a benefit.
There’s one other thing called a launchpad, which is part of the tokenomics. The launchpad is where third parties who want to launch their own token business can launch it on top of this.
Brian Rose: So you empower them to have their own token, and they use that in their own AI agent ecosystem, and it’s paired with your token.
Raghu Bala: Yes. And we give them all the tools that we have, so that they don’t need to spend the time building all this plumbing.
Brian Rose: I think the Virtuals marketplace was very ahead of the curve when it came to pairing their token with a bunch of other people’s tokens, and it gave a huge injection of liquidity that was very needed. If you try to create your own separate agent with your own token, you’re never going to be able to get the liquidity needed.
Raghu Bala: Exactly. It’s the same way as Amazon. You can create your own website or you can be a seller on Amazon. Which one gets more traffic? It’s Amazon. So that’s the same argument.
Brian Rose: This is another indication of a multiple-time founder, because you’re already thinking about the marketplace. You’re not just thinking, “Build it and they will come.” You’re thinking, “I have to collaborate and create a marketplace where there are buyers and sellers, creators of the agents and users of the agents, and make sure it’s seamless, like the App Store or the Amazon marketplace.” That’s the way to grow, and also the way to make sure the technology gets standardised.
Raghu Bala: The other thing that I failed to mention is that all of these artifacts sold in the marketplace are actually NFTs. The beautiful thing that people lost in the first iteration of NFTs, whether it’s CryptoKitties or all the digital art, is that NFTs are a great packaging mechanism. You can take anything and package it as an NFT. The benefit of having all these AI artifacts in the form of NFTs is that all of the existing marketplaces also become fair game. Your Raribles, your OpenSeas, all of those marketplaces are not alienated. They can now participate in the agent economy. We are not saying that our marketplace is the only marketplace. You create using our tools, and you can place it in multiple marketplaces.
Brian Rose: Again, that’s thinking about distribution as opposed to your own walled garden. So you could have an AI agent that’s an NFT, and it can trade anywhere.
Raghu Bala: Exactly.
59:18 Brian Rose: How does the crypto market feel for you, and how will that matter over the next 12 to 18 months? You’re based in San Francisco?
Raghu Bala: In LA, actually.
Brian Rose: In LA. How different does it feel from, say, 12 months ago? Because it’s obviously regulation and adoption, and now we have a president that talks about crypto pretty much every other day. What’s it been like for you, and are you super bullish over the next 18 to 24 months?
Raghu Bala: I think every Web3-native person has been waiting for this day. It’s been refreshing. We’ve all waited very patiently for almost ten years, 12 years in some cases, where everywhere you went there was a roadblock, and one had to set up foundations in small island republics in order to find legitimacy. I believe soon there’ll be a day when companies could launch things within America itself and not be subjected to all kinds of scrutiny or legal ramifications.
I think it’s become mainstream. Almost all the banks are using it. There’s a big movement by Amazon, Google, Fidelity, Walmart; everyone is getting on the bandwagon of either using blockchain or using stablecoins. So it’s become mainstream, and that’s a big boon. That makes things very interesting, and now I think we can start innovating.
1:02:01 The big innovation that I see happening is actually the original innovation that never took off. People are using cryptocurrency for swapping, staking and speculation, but it’s actually supposed to be programmable money. Now you can start to make your cryptocurrency programmable money, in the sense that you can put in logic. For example, say you have a transaction between two parties, and the salesperson who sold this particular asset gets a 10% commission. When the buyer pays $100, the smart contract itself will immediately kick a 10% commission this way and put 90% back into the wallet of the company that sold it.
That’s programmable money. You don’t need to wait for things to happen, whether it’s paying fees or taxes. All of those rules are built into the protocol. That’s why I see a lot of innovation happening now. People have accepted it, so let’s start to innovate, and we see tremendous opportunity in all of that.
Brian Rose: Not only could you program your money, but also give it agency. So it’s programmable money, but in a situation it makes the decision that my digital twin would make, the decision I would make, based on my history.
Raghu Bala: Exactly.
Brian Rose: That’s powerful, because you’re right. We have programmable money in smart contracts, but nobody really uses programmable money. As a person, I don’t say, “Make this transaction this way based on this, this, this and this.” Imagine how it would change my life if I just did that.
Raghu Bala: It’s so funny. My daughter took a class at UCLA, and I didn’t think of it at the time as a class on programmable money, but it was a whole book of all the ways in which money is used: interest rates, bonds, payments for all types of things. There are about 100 different formulas in it on how to program money. I was helping her with her homework, and I said, “I’m going to keep this book,” because when I looked at it, I said, “This is a blueprint for where crypto is going to go whenever it’s ready. This is what is going to happen next.”
1:04:38 Brian Rose: Excellent. You’ve had four exits. What keeps you excited? What keeps you building? Why do you still get out of bed excited to build this new piece of infrastructure?
Raghu Bala: I’ll be very transparent. In baseball parlance, my four exits have been singles and doubles. I’m looking for a grand slam home run. That keeps me going. But outside of that, growing up in Singapore, my first computer was an Apple II, and I went to a computer class where I was using the TRS-80, which is a very old computer. I’ve been coding since then, and I’m just passionate about the field.
A lot of my friends who took computer science say, “You’re still at it?” I’m like, “Yeah, I’m still at it. You guys gave up one or two generations of technology ago.” Everyone wants to make money, and that’s part of it, but that’s a means to an end for me. For me, it’s that I want to be in on the next new thing. Money will come, but let’s focus on what’s cool and let’s go and do that. That’s what keeps me driven. In that sense I’m ageless, because I like to learn new things. Honestly, I think I’ve reinvented myself five times over.
Brian Rose: Well, you’re in the most cutting-edge technology out there. You’re on the blockchain, you’re in the AI space and you’re in the agentic space, and for B2B, that is on the cutting edge. It’s a great way to stay young, because you’re surrounded by smart people and great ideas. It’s super high velocity. It’s riskier, but you can also build the most incredible technology.
Raghu Bala: Sometimes we are so in it, reading so much about it, that we don’t see it. But if you step back, like I said earlier, AI is like electricity. In the late 1800s or early 1900s, if someone knew that when electricity was invented they could go and make the bulb or the next thing that could use it, people had to think, and those things took years. Now it takes days or weeks. So you have to step back and ask, what has really happened? We have had computer programs that humans had to code for computers to understand. Now computers can understand natural language. You can talk to it, it can interpret what you said and start to even think. That’s a big deal.
I’m pausing for a second because we have to step back and see what just happened here. I no longer need to code to tell this thing something. It can understand what I’m saying in real time and start to do things. That’s a big movement from taking a programming language and coding for weeks. If you take that by itself, you can extrapolate from here to all these other things. Sometimes we are so deep into it that we miss the big picture. If you look at history, whenever there’s been a key innovation, a lot of people made a lot of money, but the ones who made money are the ones who could see that inflection point and then try to connect the dots. AI is great, generative AI is really good, and now what can you do with agents? You can go 10x from here, 100x from here.
1:08:59 Brian Rose: Right, you’re a couple of moves ahead and you see all this space in front of you. Does it bother you, as a guy who put in the hard yards as a programmer, to see these young kids vibe coding?
Raghu Bala: I have to admit I’ve tried it myself. The funny thing is that the code that’s generated is good as a framework, but when it comes to really making it production quality, I think we are still some ways off. I’ve had to debug a lot of the code it generates. But I also fear one thing. In coding there is some science and there’s some art. Elegance in coding is very important: how it’s structured, how readable and usable it is. I’m just afraid it will become a lost art.
That can happen any time you start relying on a machine. The debugging skill is very important. If you ask any good developer, they’ll say that when you debug, you are thinking like the compiler, thinking how the machine thinks. Only then can you debug properly. If you don’t know how to debug, vibe coding might take you down a path where you’re trying various things randomly without really understanding what’s going on. So I hope those listening still maintain their debugging skills. It’s very important.
Brian Rose: I think there’s always going to be a place for the purists. Right now you can go and generate an image or a short film, but if you spent ten years learning the art of filmmaking, you’re ultimately going to see it differently and be able to provide value and create something compelling, because you know all the little details. There’s that great movie called Jiro Dreams of Sushi, about a man who makes sushi in Tokyo. The whole family’s there, and one of the brothers made rice for 15 years. That was his only job, and until you make rice for 15 years, you don’t get to touch the fish. So he knows everything about rice and can master it in a way that somebody new can’t. I do think there’ll always be a place for people who really know their fields as masters.
Raghu Bala: I hope that whether it’s in moviemaking or coding or anything else, it’s a human and machine collaboration. There are pieces where we can use machines to accelerate the moviemaking process, for instance, and there are things that are still left in the hands of humans to have that craftsmanship around it. I think that mix is where the happy medium is.
Brian Rose: And that’s really using an agent effectively. You’re using an agent as a tool to help you achieve something. When I explain this to our group of investors, I always say, right now you might have a travel agent who books your travel and a personal trainer who helps you train. But in the future you’re going to have all of these agents and they’re going to work together, and create a whole brand-new ecosystem and ideas you could never imagine. That’s the way I describe what you’re doing with Synergetics. It’s a way to scale everything we do to the next level, but it has to be based on correct protocols to make sure it’s done correctly, and that looks like what you’re building.
1:13:57 Raghu Bala: Yes, and a lot of security also goes into it. We have really taken a lot of pains to ensure secure communication, secure storage and secure environments. Even the agents that we build run in a secure execution environment, like a container, so that they cannot be infected by malware. The first thing you know, some nefarious guy is going to try to attack the agent and make it into a bad agent, a rogue agent. To avoid that, we have taken about seven different initiatives.
This is one of the tests that we were put through by our partners at HPE, HP for enterprises. They really put us through the wringer in terms of testing us: what do you do here, what do you do there. I like it. They are a very good partner and helped us improve, and that’s what partners should do. Security is going to be very important. Cybersecurity, and now AI security, which is another layer on top of that.
Brian Rose: Very important. For people who are listening to us or watching us right now, how can they start using your technology, test it and find out what you’re doing, and what’s coming in the future, the token launch and so on? What’s the best way to learn more about Synergetics?
Raghu Bala: A few ways. One is that we are quite active on LinkedIn. On X we’re still a little bit on the slow side. Another way is that we conduct a lot of webinars, informational sessions, and that’s another way to get involved. Also visit our website; we put out a lot of content and blogs. You can also email us, because we have a sandbox. For those who want to test it, you can come to our website and sign up. There’s a free trial for some of the tools, so you can play with them and test them, and after the trial period there’s an upgrade path to something paid. These are important tools for enterprises to have, so do come and test out our sandbox. We’d love to hear the feedback to help us improve. And I’m also on LinkedIn; you can always reach out to me.
1:16:38 Brian Rose: I’m guessing that if you’re watching us right now and you belong to a small, medium or large enterprise, the time to start thinking about AI agents is yesterday. Your competitors are going to do this, and this tech is going to scale so fast that you’re going to be out of business, I would guess, if you haven’t been the one that adopts it.
Raghu Bala: It’s getting to that point, because every company has to move from a human-centric workforce to a hybrid workforce made up of humans and agents, and that is not going to change. That train has left the station. Now it’s a question of whether you can catch up or not. The competitive advantage of these enterprises over those who are still in the old world is not 10% or 20%. You’re talking about 10x in terms of speed, faster time to market, faster product, better customer service and lower cost. So there are going to be top-line changes and bottom-line changes.
So the investment in capex has to happen now. I’m not the only one saying this. I listen to CNBC and other channels, and that’s what’s happening everywhere, and for good reason, because the productivity improvements are significant. It’s almost the same type of movement as going from the steam turbine to electricity.
Brian Rose: And it’s happening very fast. The ones that get in are going to have that quick development, and the ones that are even a few months late could miss out on everything.
Raghu Bala: There used to be Moore’s law: every 18 months it doubles. Maybe I’m being facetious, but it looks like 18 weeks now. The AI improvement almost doubles every 18 weeks, something like that.
Brian Rose: Which is about four and a half months. I don’t think you’re far off. So get in, start adopting this, play in your sandbox and start having the conversations, because this applies to every single business out there and every single individual.
Raghu Bala: We have smaller customers, midsize and larger customers making those investments right now, and it’s improved results tremendously. One of our clients in the healthcare space used to take about four hours to do data entry of medical records. Now it takes ten minutes. It’s mind-blowing. You can’t even compare.
Brian Rose: It’s not a 10% improvement, like you said. It’s a 10x improvement, which the human mind can’t even comprehend. You’re going to show up one day and that business isn’t going to be there any more, and it’ll be too late to do something about it.
Raghu Bala: Exactly.
Brian Rose: Raghu, I really enjoyed this conversation, and it’s been exciting to get to see what you’re doing. Our whole team is super passionate about Synergetics; we have been from the start. You have that je ne sais quoi about you: all these different aspects of your personality, your education and your background, which in my opinion scream winner. We’re really pumped to be with you for the life of this project and to continue to support you with media. We’re excited about the token generation event and beyond, and it makes me really bullish on this whole industry. Thank you so much for the time.
Raghu Bala: Thank you. Thank you so much.
