Two things. Welcome back, everyone. Thanks so much for being here. My name is Rishi Jaluria. I cover software here at RBC. I'm delighted to have Dan Wagner, who is the CEO and founder of Rezolve AI. Dan, thanks so much for being here.
Yeah, pleasure to be here.
You know, look, I mean, this is a name that I think is going to be new for a lot of people in the room and on the webcast. Maybe you can just give a high-level overview of kind of the foundations of Rezolve AI. What are the problems that you're trying to solve today for your customers?
Okay, so I think it's worth giving you the context. You know, we are an AI company, and I'm not the typical 23-year-old ex-Googler who started it, so.
You don't look a day over 23. [crosstalk]
What are you talking about?
That's very kind of you to say. The reality is that, you know, I've been, you know, building businesses, tech businesses for 40 years. You know, our creds in the space of AI are as good as anybody else in that during the mid-1980s, I set up my first company, which was a pioneer in online information. We built the first commercial database to interrogate newspapers, trade journals, and periodicals. That business went on to become the world leader in online information over the next 16 years before I sold it to Thomson Reuters in 2000. During that 16-year period, apart from listing on the London Stock Market and on NASDAQ, we had to build our own search technologies. There were no search technologies in 1984. There were no e-commerce technologies in 1984, 11 years before Amazon started their business. There were no data centers in 1984.
Certainly, there were not any commercial ones. So we had to build our own data centers, build our own search technology, build our own commerce tools. As the years went on, we became more sophisticated and better at it. In the 1990s, I had two very important people working for me. One was Professor Stephen Robertson, who worked for me for a couple of years before Microsoft poached him, and he set up Bing for Microsoft. The other guy was a guy called Dr. Martin Porter. Dr. Martin Porter created the Porter-Stemming algorithm, which is the foundation of predictive typing. The concept around it and the technology around it was to ingest a dictionary into virtual memory of a computer. Then when you are typing, you know, DANC, it would guess. So it is dancer, dancing, dancehall. We use that today and have done for decades in predictive typing.
That's also the foundation of GenAI. GenAI uses that same principle of ingesting content and then using it as a reference point to guess the next word, the next word, the next sentence, the next sentence, the next paragraph to give you the impression of a conversation. Anyway, after I sold that business, I set two other companies up, which lead me to Rezolve AI. One was a commerce business called Vendor, which went on to become European market leader for enterprise e-commerce. We ran the sites in the U.S. of customers like Neiman Marcus, Land's End, TJX Companies, Under Armour, J. Crew, LL Bean, and many others, mid-market retailers. That business was bought by Oracle in 2014 as their main commerce platform.
The other business was a search company called Attract, which listed on the London Stock Market in 2014, sold to private equity in 2019. The reason I give you all that background is so that when I started looking at the developments in GenAI in sort of 2013, 2014, before it became something that we could build on, we knew it was coming. My technicians were working in the open-source community contributing to its development. When the transformer algorithms, which are the foundation of GenAI, became available in 2017, we were already all over it. We knew that it had one big problem: hallucinations. We wanted to solve the problem of hallucinations.
The use case we wanted to create for Rezolve AI was to address the problem of checkout attrition and cart abandonment in e-commerce, something we knew a great deal about having run the e-commerce platforms for all these major retailers. The issue is really around the fact that seven out of 10 people who arrive at an e-commerce platform or a digital platform do not end up buying anything. Seven out of 10 people who walk into a physical store do end up buying something. Why is there this huge disparity? For decades, e-commerce directors and marketing directors have been trying to solve this problem. Partly, it's been down to, can we solve the checkout flow, make it faster? Of course, that's happened with Apple Pay, Google Pay, Stripe, and Autofill and things like that, which have made it easier to check out.
That has made no difference to this attrition. What about surfacing up shipping or making it free shipping? Made no difference to attrition. What is the problem? We believe we know what the problem is. The problem is a lack of information about the products and services that you are buying. If I was to ask my wife to buy me a mobile phone, she could not do it online because she does not know what a megapixel is or a megabyte or an OLED screen or iOS or Android. These phones are presented in that way on the web. That is how they are shown to me. I could not buy it, right? If she went into the AT&T store on Lexington and 34th Street and said to the sales, "I need to buy my husband a mobile phone," the salesperson would say, "Does he use a PC or a Mac?
Does he use his phone all day long? Does he need long battery life? Does he need a good camera? A few questions. Here are the phones that you should get for him. One of these phones, you'd buy a phone. What we're trying to do with Rezolve AI, obviously solve the hallucinations, but actually provide an interface and an experience online that is akin to dealing with the world's best salesman in a physical store. That is what we've done. It's really absolutely amazing. It's fantastic. I could demo it, but I'm not sure I should right here, but I could. It's incredibly impressive. Maybe I should. I spoke Mike. If you're interested, I will. The point is that we believe that the existing e-commerce and digital channels are end of life.
Holding on on a call center for hours and dealing with somebody who doesn't have the answers to your questions is end of life. Going onto a website and then having to navigate around a structure and a thing and search and get 40 items, 40 items is end of life. It's all finished. It's over. What we're doing is we're completely changing that to a whole new way of interacting, which is really fantastic.
Yeah, that sounds great. Maybe let's talk about something that you brought up during that, which is that you've been able to basically solve for hallucinations. Can you walk us through how you've been able to do that and what has put you in that advantageous position when clearly it's elusive for a lot of AI companies out there even today?
Okay, so first of all, the reason it's elusive to all the other AI companies out there today is because the people who built it were not considering, you know, there's a kind of technicians approach something by saying, "Oh, there's this new capability. It's going to solve everything." They kind of go blindly down the path of backing it to solve everything, which sometimes doesn't work like that. That's partly the reason that everybody else got away. Of course, now they're all backtracking and building guardrails and having AI oversight and all sorts of things to stop what's wrong foundationally with the process. I come, as I explained before, from a very, very long history of managing and structuring data. I was involved in the early stages of this moving to probabilistic algorithms. Now, I mentioned Dr.
Martin Porter and his probabilistic algorithm for predictive typing. Before that, what we've been using to search Google is binary mathematical algorithms. Binary mathematical algorithms is two apples and two apples is four apples. There is no hallucination in that. You cannot have a hallucination in a binary mathematical algorithm, okay? Probabilistic algorithm says two apples and two apples equals water because each apple is 60% water, 20% fiber, 20% sugar. So the probabilistic output of four apples is water. Now, I'm being deliberately, you know, antagonistic with that idea because it probably wouldn't do that. But probabilistic algorithms, by their very nature, make mistakes, right? They make some judgments that are wrong. Now, if you take a clean probabilistic algorithm, it's going to make mistakes 3%-4% of the time.
If you then make, you know, confuse that with a load of data that's inaccurate, poor data, and the internet is 40% rubbish and social media is 80% rubbish, if you ingest everything on the internet and ingest everything in social media, then you are going to exacerbate that error rate dramatically, okay? Which is why you have these problems. We approached it with the, you asked the question, I'm trying to, it's not an easy question to answer. We have three patents around the way in which we solve this problem. It's taken us nine years. The first is that when we ingest a product catalog, we create a very rich taxonomy around it. A taxonomy is when you catalog the item with tags, okay? We create an AI-generated taxonomy that's mad. You know, the product catalog might be this big.
The taxonomy is enormous around it. So it will take a running shoe and it will say, "That's 100% a running shoe, 70% a trail shoe, 50% a casual shoe, 2% a bar mitzvah shoe, 3% a wedding shoe, 5% a date night shoe," okay? Just to give you the idea. We tag it to the nth degree with what it could be appropriate for that shoe, what could it be appropriate for. We use AI to do that. That's patent number one. Patent number two is that we take a query. I'm looking for a casual shoe to wear at a friend's wedding. I'd like it to be a white shoe to go with my blue suit. That's the kind of question I might ask. We then decompile that question into what's known as semantic search terms: white, shoe, casual, wedding, men's.
We throw that at this enhanced catalog and we get a very accurate, because there's no hallucinations with binary searches, accurate set of results. Now we've got this set of results that there is no hallucination with, right? It's accurate. It will be ranked in terms of the most relevant running shoe, you know, casual shoe for a wedding down to the one that's the least, but it will be all of it relevant. We take, and then our AI steps in again, looks at the original query, looks at the data set, and answers the query using natural language. Can't hallucinate. When I showed this to Marcello Bonatti, who's head of AI at Microsoft, he said it was absolutely brilliant, which it is, by the way. It's a very simple, I'm not a technician.
It's a simple, elegant solution where we use AI where it needs to be used and we don't use AI where it shouldn't be used today. That's how we solve the problem.
Yeah, super helpful. Maybe can you walk us through, you know, where you're seeing the opportunity with agentic commerce, right? Because clearly it feels like with this technology, there's a lot more you can do, but that's clearly the big use case today. Why do you think the technology is suited to this specific use case? And you know, I know recently Andy Jassy on the Amazon call talked about the opportunity with agentic commerce. So maybe walk us through that.
I think first of all, people need to understand what agentic commerce is. I think, you know, it's a term that's being bandied around. It's not really a great deal of understanding. Agentic commerce means that an agent is making the commerce, is interacting with a website and making the decisions on behalf of some instructions that have been given maybe by you, but you're not on the website. So you're not going www.macys.com and then interacting with the website. You are interacting maybe with Siri or with ChatGPT or with Google Assistant or something, right? And it is going off and talking to Macy's on your behalf, okay? That is agentic commerce. Now there is great opportunity here and also a huge amount of pitfalls.
The opportunity is that for those retailers that provide an agentic interface, they have to build a doorway for this agent to come into to find out about what's available and what the stocks and everything else. The opportunity is that they expose themselves to this broad audience of Siri users and other people. If they don't do it, they're not going to get a look in, okay? That's on the plus side. On the negative side, it's like opening their warehouse doors to a bunch of lunatics because there's no control. They're opening their doors to their e-commerce site, the back door basically to their warehouse, and allowing these agents to come in and take stuff off the shelves and sell them to people over here, right?
The problem is that they're going to end up, and those, as we just described, those language models, those technologies that are being used to come in through this back door and interrogate their e-commerce catalogs are hallucinating all the time. There is a very high likelihood that I'm a consumer. I want to buy a pair of trainers. I end up interacting with ChatGPT, but it's going off to Adidas and Puma and so on. It ends up buying me a pair of Adidas shoes, but it's the wrong color, it's the wrong size, or it wasn't in stock, or it wasn't right for the purpose I was asking about, or something because of the hallucination. I go back to Adidas and say, "I bought these shoes and they're wrong," and they're going to have a big problem.
Mark my words, two years, three years from now, this is going to be a massive problem. There are a lot of retailers that are going into this without any clue what they're doing, and they're going to open the doors their way. It's all going to be a mess, okay? What we've done is we, as part of our solution, the first thing we do is that if you go to www.adidas.com, you end up having a conversation with a very knowledgeable salesperson, and you end up buying the right shoe. That's solution one, okay? Solution two is we're providing an agentic interface to the agents, but we're not allowing them to come into the store and warehouse and take stuff off the shelves. We're answering their questions the same way that we would answer the question of the consumer.
The agent comes in and says, "Do you have these in stock?" Instead of going in and looking itself, it has to ask our bouncer who's standing on the door of the warehouse, and our bouncer says, "Yeah, I've got this in stock, and it's available in this color," and so on. If it knows, our agent knows ultimately who the customer is. It says, "Oh, Dan Wagner's a regular customer. We'll give him, you know, a special, you know, polished thing to go with his order or we'll upgrade him, or we knew he ordered this previously. We'll try and upsell him that." That control is still being maintained by the merchant. Back in the noughties, at the beginning of the dot-com boom, Amazon encouraged Toys R Us to put all their products on their site.
Toys R Us lost control of the customer, went out of business. Everybody who rushes into this without providing the right controls to the agentic interface is going to do the same thing. We are providing a solution that solves both problems.
Yeah, makes a lot of sense. You've highlighted these partnerships that you have with Microsoft and Google. Can you talk a little bit about what they're contributing today and how do you see those partnerships and maybe future partnerships evolving over time?
We all know how data centers, you know, and, you know, cloud computing data centers for this new highly compute-heavy world that we're moving into is a very well-funded, contested market. The dominant players at the moment in the West are AWS, Google Cloud, and Microsoft Azure. Now, those guys are also investing many billions or tens of billions of dollars in building out their infrastructure to support this growing wave of demand. There are other players coming into the market, like SoftBank with an open hour with a $500 billion investment and many, many others who are also coming into the market. There's going to be a lot of greater levels of competition than the three incumbents in the West previously.
As a result of that, you know, there will be a lot of movement when contracts come to an end, people moving off Azure to some new player who can do basically the same job. What Microsoft and Google have recognized is, one, we have a solution that's fit for purpose, can be sold through their industry channels, and they're doing a very good job helping us sell through those industry channels, but also can provide a very sticky solution to their core cloud services. Because if a retailer who's on Azure buys Rezolve and, you know, deploys it, then it's more difficult for them to up sticks and go to another provider, you know, because all their data, all their cloud, all of this stuff is in there.
The way that it works with both these accounts, both these partners, which goes right up to the top level of both organizations, is that they are promoting Rezolve to their customers. They're encouraging their customers to buy Rezolve by giving them dollar-for-dollar credit against their commitments to Microsoft and Google. These companies typically have multi-year commitments, and they're saying if you buy Rezolve , it comes off that commitment dollar-for-dollar, which is great for us because when we present it and people like it, there is no budget discussion because they've already allocated the budget to Microsoft or Google. They're also incentivizing their salespeople dollar-for-dollar that if they sell Rezolve , it counts 100% towards their quota.
We've got this wonderful dynamic pincer movement, and they're doing it because they recognize that we provide that stickiness and that, you know, from a corporate perspective, outside of, there's nothing they're selling into the corporate market. They're saying, "Here's a language model, go and build something." I mean, if you just dig a little bit into this, you'll see AI is not being used by industry yet. It doesn't work because it hallucinates. Can't leave things without oversight. The deployments of any AI in business today has human oversight. Like, for example, in supply chains, the AI might recommend something, but a human being will say, "Okay, buy that, buy these additional items." You're not letting the AI completely run amok.
That's because it doesn't do it, apart from with us, where you can put it front of center into your website and let it serve your customers.
Yeah, okay. That makes a lot of sense. Maybe can you talk a little bit about your go-to-market motion, how you think about, you know, the interplay between both product-led and sales-led growth for Rezolve ?
Yeah, so there's three channels that we're going to exploit to capitalize on to become a market leader in the space. As I mentioned before, I built some market-leading companies. It's not my 1st rodeo, but the 1st is obviously through organic sales and building out a sales infrastructure. We haven't really been properly capitalized. We came through a SPAC, terrible idea, worst moment of my worst few years of my life. Came out of that, had a lot of liabilities, had to clean that up. As we came into 2025, we were finally released from everything, were able to focus on driving the business. It's only very recently that we've been given significant capital to do that. A couple of hundred million was invested in the business a few months ago, and we're now able to attract really great talent from industry.
Crispin Lowry is a recent appointment from Microsoft, a very senior executive at Microsoft. He is rolling, he's now Chief Revenue Officer at Rezolve . He is rolling out a 40-60 person team in the United States, a 20-40 person team in Europe on the street, supported by marketing, all happening in 2026. We expect to be profitable at the end of Q2 2026. That channel is just really building up. We've only had 22 salespeople, and we've gone from zero to the end of this year, we exit with $150 million ARR. We've done amazingly well with a tiny amount of people because we've got a great product and great partners. We want to get out on the street, and we want to sell directly because that's always the most productive.
The 2nd channel is Microsoft and Google, who are giving us leads, and that's going great. I'm not stopping that. Obviously, that's going really, really well. We'd like to add other partners to that. The 3rd channel is M&A. You cannot become market leader through organic and partnerships alone. You can't do it. You have to acquire businesses because I can't be in Spain as well as opening up North America. I can't be in, you know, Italy and, you know, Mexico and so on. Acquisitions give us a footprint. They give us talent. They give us customers, and they give us the ability to upsell in those markets. We see all of these three channels as being the way that we can fast execute the land grab. We're not a big R&D AI business. We're not spending billions in AI.
Our burn is $4 million a month at the moment dollars. As I say, we expect to go profitable in Q2. We built our product and our solutions. Our deployment is very cost-effective. We're in sales and growth mode now.
Yep, yep. Maybe let's hit on the metrics. Sorry. So you talked about you're going to exit this year at $150 million ARR, and if my numbers are correct, $500 million, you said next year.
Correct.
What are kind of the building blocks to get to these milestones? What's giving you the confidence in these sort of growth rates?
I don't think anybody believes that, right? Because we did the first half of the year, we did $6.3 million in revenue versus analysts' expectations of around $5 million. People are really scratching their heads, you know, how are they going to do it? They're going to find out in March, right, when we announce our full year results. I'll probably give some updates, but until they're actually, you know, audited and out there, I think the market's a bit skeptical, but that's fine. I've been telling the market, giving them visibility. It's not like I've been saying it. They'll find out then. In the meantime, it's a great opportunity to come into the stock because I think we're, you know, heavily discounted for that very reason. You know, we don't need any money. We're not in a, you know, we're fine.
There's no pressure on us to worry about the price right now. The investors who came in, the $200 million recently raised at $5.40, and it's about $2.75. I don't know what it is today, $2.80. You know, they're not concerned because they didn't buy into it for a short-term gain. They believe in the vision as I do. I'm the larger shareholder. You know, as the next 18 months unfolds, you'll see a very, very different picture emerging.
Yeah. And maybe just in terms of like even just the building blocks, right? I totally get, you know, you're saying market doesn't really appreciate this, but you put these numbers out there, you clearly have confidence. What is it that's giving you confidence? Is that based on like pipeline? [crosstalk]
Pure visibility.
Okay.
Pure visibility and execution.
Yeah.
You know, we're seeing it. And also experience, I guess. I mean, I also know how damaging it can be to say these things and not meet them, right? I mean, you know, that's another factor, of course. I wouldn't be saying it unless I'd be confident, so.
Of course, of course.
You know, we upgraded our numbers twice this year. We entered the year saying that we'd do $100 million ARR. Mid-year, we said we'd already done $70 million. We'd already committed contracts that would exit the year with $70 million ARR. I've never been in a business in my career that's had this kind of takeoff, but I'm very comfortable with it. My team are less so. They're all running around like madmen, but it's a good dynamic. I mean, you know, we're onto a winner here. It's great.
Yep. Yep. No, that's super helpful. Maybe, you know, what can you share that you're seeing as you talk to your customers and look at your customer base and prospects in terms of just the state of enterprise AI demand today? You know, as you have customers worrying about data privacy, security, et cetera, you kind of alluded to this a little bit earlier. What can you do to help get over those hurdles for those customers?
I think we do. I mean, I think, first of all, I think we come across as confident and knowledgeable and experienced. I think what they hear from others is a lot of hand waving and, you know, vaporware nonsense, you know. There is a lot of, you know, oh, you can build this on that, and here is a language model, and we can help build it for you and spend a few million dollars with us, and we will build it for you. We are saying, here is a product. Here it is. Let me show it to you. Here is what our other customers are saying, how it improves their engagement with customers. These are things you need, and we are the partner that you need to come with us.
You know, that's endorsed by Microsoft sitting beside us saying the same thing, or Google, because whoever the customer is, they're likely to be Microsoft and Google. If I was to come in organically to a retailer, one of the first things we say is, who are you hosting with? They'll say it's either Microsoft or Google, because very few of them are with AWS because they're competition. Once we know it's a Microsoft or a Google account, we call Microsoft and say, who's the account for this one? They come in and help support the sale. We've got a lot of good kind of momentum as a result of that. That's driving a lot of this fast takeoff.
Yeah, yeah. That makes a lot of sense. You know, given the size of the TAM, I'm sure you're running into competition. Who are you seeing that you're competing with the most often? How do you think about your own right to win, especially as there's new companies coming up every day?
Look, I do not want to sound, you know, either ignorant or arrogant, but in both cases, you know, we are not seeing competition that provides a solution like this, where there is a lot of people talking about things, but there is nobody doing anything. There is a lot of people packaging old things and saying it is AI and chatbots and rubbish like that. You know, there is nothing out there. I actually want strong competition because it is a massive market, and competition is a healthy stimulant to the market as a whole. It drives business for us because if something else is out there promoting, a lot of companies will say, who else is out there? Let me do an analysis, and they see us, and then they invite us in for a pitch, and we will win. I have seen that before in previous businesses.
I kind of want to encourage proper competition, but it's not there yet. Will it be there? Of course it will. It's only a matter of time. The big problem is this hallucinations right now. If you have hallucinations, if you're building on OpenAI or any of those models, you're going to have hallucinations. You have to build your own model. Building your own model is an expensive task. Took us nine years, cost us $130 million, and that's extremely cheap. It's a lot of money, but it's extremely cheap. We knew what we were doing, and we had all the right people, and we were right in on the ground floor.
I think that somebody coming in to take us on needs to have a lot of sophisticated abilities and be willing to focus on this vertical when there's lots of stuff out there that they could be doing where they don't have us in place. I think we've got a clear run for them for the next few years.
Yeah, yeah. Maybe look, Dan, you've been through a lot of technological shifts, as have many of us in this room. Maybe can you talk a little bit about, you know, as we think about this shift with AI from your traditional pre-packaged SaaS applications to multi-agentic systems, right? And maybe even the way we interact with them, going from this GUI paradigm we've been in for 30 years to more natural language. How do you see this whole landscape and the way the surface area we interact with all of this evolving? And how do you think about your own kind of positioning to win in that new paradigm?
No, I'm not the only person to have said this. You've heard it a million times from everybody who's anywhere close to the space. What's happening with AI is the most transformative thing we've ever seen. If you're in technology, you've been in technology, it's the most extraordinary change. You've got to think about everything that we're doing right now is like, that's done. The future is about just having natural language engagement with computers in a way that we're only just seeing the beginning of. You know, we've been trained to go www. somewhere or other and then navigate and click on things. They're so unnatural. It's not a way.
Whereas if we go to the store, whatever that store is, we talk to somebody and we say, I'm looking at this, and that's the experience that is now available, is now possible in these digital channels. The moment that happens, the moment it changes everything on the demo that there's a video of it and so on that we show is that, you know, I could say to a hotel site, you know, do you have, you know, I've got an allergy with feathers. Do you have the appropriate bedding? I like to bring my dog to the room. You know, how close are you to Central Park? Do you have parking, you know, nearby park? Questions that you can't answer today. I can't answer those questions if I'm booking a room on a hotel. I have to call them up.
Do you have appropriate bedding for people who've got allergies? Or do you have, you know, can I get waffles in the morning sent to the room? Or I have to look for the room service menu? You know, I can just get those answers by asking the question. That is a massive change in the way we interact with digital platforms. We are leading this. We are ahead of it. We have got a product in market. I think that we are very well placed to capitalize on that. I do think that what we are seeing today is just over. It is like if I took you back to 1994, we would all be really horrified at how we would have to go about our daily lives because we do not have the internet.
We can't message people in an instant on our mobile phones and get answers to things that we would otherwise. We'd have to go back to the old way. The old way was painful. Going to the library and I don't know what, right? We had to do all sorts of things that you had to do back then that we've forgotten about because we've moved to a much better way of doing things. This shift is that on steroids.
Yeah. Okay. Maybe with the last 45 seconds we have left, I mean, you've talked about some, you know, potential use cases. What's a current use case with a customer that you can talk about where you're seeing them use Rezolve? And how are they measuring their kind of success with it?
A good example is Dunkin', Dunkin' Donuts. We're on their app for their customers. I'm driving along. The Dunkin' two miles down the road. I'm driving along. I go past a Starbucks, a geozone, which we, as part of our solution, triggers the app to wake up and say, "Hey, Dan, would you like a donut and a coffee this morning?" I'm driving along. I go, "Yeah, that'd be great." I say, "Well, can I get you anything else with that?" "No, thank you." I'm driving, right? It says, "Can I place that order for you?" You can pick it up two miles down the road on the right through the drive-thru. I go, "That would be great." At that moment, at checkout moment, it appears on the screens.
We're embedded on the till screens at Dunkin' saying, "Dan Wagner ordered a donut and a coffee, and he's five minutes away, four minutes away, three minutes away, two minutes away." When I drive to the drive-thru, it says, "Welcome, Dan." Knows exactly within 25 ft. "Welcome, Dan. Your order's ready. Just drive around to the pickup window. Go around to the pickup window, pick up, and I'm out." Five minutes saved through that checkout flow for Dunkin'. Much better experience for me. They stop me from going into Starbucks because the geozone triggered that engagement. That's one example.
Awesome. I think it's a great place to jump off. Dan, thanks so much for being here.
Pleasure. [inaudible]
Thank you. [inaudible]