Is a webcast or audio video? An audio webcast. There's no video. Okay. There's no video. Okay. Thank you, everyone. My name is Tom Singlehurst. I actually am head of the European Media Team within Equity Research at Citi, but I also look after Global Education. Before we start, a couple of prior notices. There are some disclosures available. If you can't see them, either you know, either on the webcast or at the front desk, please let me know. It would be a pleasure to send them through. Apologies, we're running a little bit late. Harish fell victim to the lifts, which is a common theme here at the Citi TMT Conference. Wonderful hotel, but it's difficult to get around. Harish, it's obviously a huge pleasure to welcome you to the Citi TMT Conference. Before we get going with the conversation, I know you wanted to say a few words about recent developments. So maybe before we begin sort of the fireside chat, like, you know, should I hand the floor to you just to? Yeah, sure. Thank you, everyone, for being here. A real pleasure to, you know, come and speak to all of you. I know there are a lot of people listening in. So I just want to start over by saying there was a short seller report that came out last week that was really filled with, you know, in my mind, many misleading statements, false assertions, and across the board. And, you know, I think for us, what we have done as a company now is really established a process to go through this to address each of these allegations. And, you know, and I really believe all of these allegations are false. And I think we're going to have a, you know, a very strong response here. So you know, there are a few things that was said in here in that report I just kind of wanted to bring in. You know, it started the premise of the whole thing was, you know, the company is not real, no products, no customers, no revenue, or something on those lines. You know, we've always believed that, you know, we have probably a best of class, best of breed AI platform and industry. You know, we saw certain things in this industry well before others did. You know, but really, we feel very comfortable saying we were and are one of the pioneers in this space. I just want to touch on three things, and then we can just jump into our discussions. The first thing was we recognized back in the day, when we built our platform, that there was a need for training data. I just believed from the beginning that training data is very different than other kinds of data, and that you had to pay the owners of the data for that, or license from that. This was back in twenty eighteen, and we've had people across the board ask us, "Why would you do that?" More recently, if you look at everything that's going on in AI, that has become the number one issue, and everyone's getting into data and content licensing contract. I like to think that if you see something six years ago, then we should get some credit for that. Number one. Number two, we saw well before all these others that for AI to be effective inside the enterprise, it had to be on-prem. And to that end, we built our enterprise models that were a combination of narrow models and smaller models and bigger models. And we have over 40 plus enterprise models that on which many, many use cases can be built, that are designed to work on-prem, meaning the unit economics for those on-prem installations would actually be, will work. So that was the really second critical piece. And really, the third thing I just want to bring in is a challenge for every enterprise has always been use cases, and how do you build a use case in a very effective and cost-effective and efficient way? And we saw that one, too. So I think that's a really important part of what we do. And the other allegation that was made was about our customers. And, you know, and I think one thing I want to say, that we have for the last four years or more, I mean, I think we've had a very strong, strong track record of growth. You know, over $400 million in revenue in 2023, and now the last audit report, we have four years of audits. There's a lot of work that we have done, and, you know, we've been funded with a very limited amount of equity, but we've taken on a lot of debt to this. We have two lenders from whom we borrowed over time, and, you know, I think it's... When it comes to lenders, what they first and foremost want is paying customers, you know, not as much, so they don't participate in the upside. What they want to focus on reducing the downside. And so we have paying customers and, you know, it's a company that's been, you know, best executed teams that we have. We also took oversight from the beginning very seriously. We put together a world-class board. And so I think it's really just important to understand. And for us, it has been a little unfortunate that we had to deal with this. You know, and during this conversation, I think we'll talk more about the journey and everything else, but I just kind of wanted to put that out there. Okay. Well, that's very clear, and as I suppose, you know, launching into that discussion, I suppose one of the great benefits of the de-SPAC process, but also one of its challenges, is it's so sudden. It happened so quickly. So, you know, you don't necessarily get a lot, the market doesn't get a lot of time to get to know a company, and understand what it does. So maybe we can do that and go back to the beginning. I mean, you've given us a decent amount of background there, but, you know, for you, what are the sort of main milestones since, you know, over the last 13 years you've been up and running. What are the main milestones that for you, for the business that have brought you to where we are today? And then we could maybe drill down into some of the use cases and... Yeah, no, absolutely. My background is I started out in Silicon Valley at a company called Sun Microsystems, where I was a microprocessor architect, and I worked on some of the most successful products there, and I think I built a really good reputation among my peers there in terms of buildings. And, you know, it's really that background that I brought to the, you know, microprocessor is about building very high-performance systems, and we took that to other industries like telecom and other places. And, you know, I also was part of a startup, but I think, you know, we raised a lot of capital for the startup, and then, you know, I think we just didn't. We arrived the traditional way, where we hired a lot of people, and we got caught up, and things didn't work out. I went to business school at Stanford after that, and when I graduated, I just told myself, I think, "You know, when I'm raising money to build a startup, we're gonna do it very right and very capital efficiently." So, you know, when we started iLearning back in 2010, early 2011, I got a limited amount of money to raise from you know people who I knew who were interested in this, and it was important for me to make that count. So capital efficiency was the key for me. And I also knew how to build large-scale, robust, scalable systems. And so for me, when you build a product of any scale, you need a very large team. So we're going to have times where you start building the product, we need 100, 200 engineers, and then you got to wind it down to five or 10. That's not a very feasible way to hire a... You know, a startup that hires, reduces 80% of its workforce has no chance of surviving. So the only way you can do it is to work with a partner who can help you build that technology, who would allow you to ramp your team up and ramp your teams down. And so, so from our standpoint, we relied on working with a really good technology partner to, to be able to build our various different products. And what... These technology partners also have other customers, and so, you know, we were, from my standpoint, working with some of these partners, able to get back into these customers' hands. There was a lot of success. We were able to get good early revenue, and that time, we had to start building new features, and my choice was go and raise more equity. A lot of our pro customers were multi-year, had multi-year contracts. And so our partner told us, "Listen, we can build the solution for you, and we can collect from the customers. You know, since we haven't used them, we can start that." And so from my point, I thought that was a great thing. So then I said, "Listen, the most important thing for me is IP, the company. Mm-hmm. And, you know, so we were able to get into the space where we had a debt from a partner that was secured basically against the receivables, and that's how we built this product. And then, you know, we were able to rely extremely again outsourcing support and service. So when you sell to customers, and once you sell, they want to know how are you going to support that. And so and then all that, at that point, we've been trying to phase and start to build more and more operations in, and we come in. Subsequent to that, we had. So through that process, we were able to raise money from the tech investor, and that's kinda and then we were able to borrow additional debt later. That's kind of brought us to where we are and what our plan with going public here. Because along with working, it wasn't really the choice of going SPAC versus IPO. It was about working with a firm that would help us bring that operational capability for us. I think PE firms are generally really, really good with that. That was one of the key drivers for PE. With the SPAC, it was a PE-backed SPAC. We built that. We've been on this journey to kind of get things to, you know, getting more ship in-house, building more of those capabilities, and we've done so. Right now, as part of our agreement with our lenders, all our customers had to pay into our account. We made this transition. It's already there. I think sort of this has always been part of the journey, and I think, you know, when it comes to a SPAC, I think, and one of the things I think in this process, we were very much deep on just operating as executor, and one place where I think we didn't do as good a job as I think we need to do is, we didn't put ourselves out there and participate in the thought leadership discussions, so when you see things early and you don't participate, and this is my learning that I've learned just over the last few days reflecting on things. When you see things and do things early, and if you don't participate in the thought leadership, then people don't quite understand why you did those things. Mm-hmm. And that was really the part of it. But, you know, we're excited. I mean, I think we feel like we have a best-in-class product in AI, and I'd love to talk more about- Yeah ... how we got here, and then- Maybe we can do that. Yeah. I mean, actually, one of the things I think would be useful, certainly for me, is just to sort of recap on the sort of use cases of the iLearning proposition. You know, people looking at the business for the first time, what challenges are your customers trying to solve, and ultimately, how do you solve it? Right. Okay, we are an enterprise AI platform for learning and work automation. Mm-hmm. Companies deploy our platform, and on that, they build various learning automation use cases, work automation use cases, et cetera. Almost every company today is looking to understand what AI means for their enterprise and how can they bring AI into their enterprise in a, you know, in a safe, responsible, ROI-centric way. The challenge for most companies have been AI is one of those things that has come at them so fast, there is a significant fear of missing out across all levels of an organization. And so they're really trying to figure out, you know, what to do. And I think the proposition present to them today is you must spend millions of dollars on something with this promise of future ROI, otherwise you'll be left behind. I think that's really the core area where we are helping organizations, you know, bring AI into the enterprise. And what are the factors would you say that really differentiate what you do? Is it, is it the fact that you... I know, I've reading the materials, I know you say you have this out-of-the-box solution. Is it the speed of deployment? Is it the cost? Is it the outcomes? What would you say is, or is simply no one else doing it? No, no, I think. So we've spent several years building our AI platform. So if I have to really summarize what we are, is we are a low-code AI platform for learning and work automation. And so what this really means is we are a platform on which people can use low-code infrastructure to build these use cases in a matter of weeks to months, at a price point of maybe in the hundreds of thousands, versus the dimension, the most common thing today, which is building custom solutions on top of hyperscalers, etc., that requires expensive AI engineers, and that takes years, and so this doesn't scale, right? That's number one. I think the number two that we have done, like I said, is on-prem and data sovereignty is a very critical thing for enterprise. So how do you make an on-prem installation of AI happen? And that comes with building these models that don't consume a lot of, you know, compute, that have good unit economics. The third thing is really verticalization at scale. We have these enterprise models for specific verticals. So if you think about insurance as an example, right? We have models for that vertical, and then on that, several use cases can be built, that might be already pre-built. So this would be like, for example, claims intake, claims processing, early notification of loss, as an example. So that out-of-the-box platform is then given to insurance. That vertical, everybody knows this. You probably most enterprises know the need for that, and so that's the critical part of it. So I think those are some of the really key differentiators. So in other words, having the ability to build these use cases in a very efficient way, being able to support, you know, an on-prem. In this case, still do the cloud, but being able to support that on-prem infrastructure, I think is a really critical part of it. And the third thing is really the specialized data that we have. You know, because each of these industry verticals requires very specialized data. And, you know, we've been partially, I think, then buying that specialized data. So, right. We'll come back and talk about data. Can we talk a little bit about the evolution of the sort of customer journey? I mean, what does a sort of typical sales process look like for you? How long does it... I mean, who are you talking to within the organization? How long does it take to sort of convert clients? Right, so we are in, like I said, we are in several verticals. We are in 12 verticals. We always taken to heart this verticalization of scale, strategy. Mm-hmm. So, our sales process is typically what we call defined POC driven. So in other words, we start with the traditional process, where you go through prospecting clients, you do demos, and then we get into a proposal stage with the customer in terms of what the use case might be. And they have the established budgets. And then based on that, we will do a design, a proof of concept for, say, three to six months. So they get to see the system working in their world. And our conversion rates have been pretty strong, you know, like in the 70% range. So I think that's been a big plus. But you got to go back to... And I kind of look at the AI world as the pre-ChatGPT world and the post-ChatGPT world. Imagine selling AI in 2018, 2019, 2020, in a world where people are like, AI, they think of AI, AI as this futuristic technology. It was really important that people got to see the platform working, you know? The second thing I think was really important for us was building with use cases from the beginning. You know, today we have this unique case where you have all this Gen AI installation, and everybody's going searching for use cases. "I need a use case." And when you sell AI in a world where no one knows AI, you had to start with a use case. That's where we started out. The first use case was around learning automation. And we, that was one of the categories we sort of, I feel, pioneered, was the idea that, hey, don't worry about the fact this is AI. It's still learning. But what we're going to show you is a software system through which you can scale the footprint of learning in an organization. Now, here's AI, through which you can automate some of these functions. And so people really understood existing use cases, and so this was just an upgrade of their existing infrastructure. While we're talking about sales, within your commentary, you've talked about using value-added resellers. I think there was 30 at the last count. Can you just recap on why you use that approach and what the sort of upsides and downsides are of that? Right. I think, you know, AI is a groundbreaking technology, you know, almost like electricity. And I think there's a lot of people who talk about how AI is the new electricity, period, you know? And, I think it was the founder of Android who said that. I think that was one of the most well said statements, right? So imagine a time where the first producer of electricity started going and knocking on businesses' doors and saying, "Hey, would you like to buy a few kilowatt hours of electricity and build appliances on top of electricity?"... and enterprises don't know how to do that. Okay, they. And so instead, they went and partnered with these appliance makers, whether it is refrigeration systems, heating systems, go back to them, right? And so these value-added resellers are the appliance makers of AI. You know, with the same analogy, the idea that I can go to an end user customer and tell, "Hey, I want you to build a solution on AI," is, like I said, just as hard as telling them to build a new appliance. And so that's why we, from the beginning, we saw the need for doing it through value-added resellers. And so for us, that I think is a for any AI company, that's a very important element. And we call these appliances AI apps, or we call them AI engines. And so what we have been doing is building a marketplace of these AI apps for each vertical. And also, so these value-added resellers are using our platform to build this, as well, a lot of those. They then sell it to their end customer, right, and so they understand that well, so you know, when we go into an insurance role, let's say, claims intake or early notification of loss of a system, the conversations around that are not around the capabilities of the AI platform, and 45 into the customer, you're not really talking about AI, you're only talking about that product, and these guys are really, really good with that, and I think that's how we've always done it, and that's, I think, is a critical part of our success. Again, you know, So yeah, they're providing the use case, and you're providing the extractive AI. Foundation, the platform. The underpinning. And the tools to build those AI systems. So traditionally in the past, to be a solution maker, you needed to have very strong programming skills because that's how it was. But I think, with this low-code, you can bring other people into the network. So you could have partners, not just be the SI, but they could be aggregators of, you know, hospital systems, or they could be content providers and so on. So it brings up a whole new set of- So if we go back to that customer journey, one of the resellers will, you know, come up with a use case. It'll be sort of powered by your technology. They will demonstrate proof of concept. It'll get embedded. Once the relationship is up and running, what are the sort of milestones for customer success, and how do you grow? I mean, is growth really just multiple additional clients, or is there a land and expand? How does... What's the growth trajectory look like? So our one of the biggest metrics that we use is what we call net dollar retention. Mm-hmm. And our net dollar retention has been between 115 - 130+ over the last several years. In fact, Q2, our net dollar retention- Mm-hmm ... 130. And so I think upselling is critical in this. So once somebody builds this, providing them all the support they need to upsell and sell is a very critical part of what we do. So really, growth for us is upselling, adding new applications, and adding new customers. They're all. It's a combination of all those things that one would see. Perfect. And that's, once again, that's still conducted via that sort of value-added reseller. They'll be managing the, you know, the- I think it could be both. So we work very closely with them. So when we go into a vertical that we haven't been before, we would identify a value-added reseller who brings that domain expertise, and we work with them to build our enterprise model, because they know the landscape really, really well, right? And maybe the first initial few use cases, we would pull it in, and then we would help them build other use cases. So they bring a lot of expertise upfront. And then once this is built and selling, we do a lot of, you know, like marketing events, where they invite their end customers. We come in there, we talk, we get people to sign up to these things. And that's really how this process works. But at the end of the day, they are best positioned to sell the solutions to their customers than we are. You know, we just focus on helping them with some of these things. But yeah. Perfect. I think you've really highlighted education, healthcare, and areas like insurance as industries that have sort of really embraced the offering. What is it about these industries that make your tools so relevant? So I think, you know, the use cases come in all shapes and sizes, and I think what we found in these verticals is these use cases can build multi-billion-dollar market on their own. So if you take something like insurance, and you have early notification of loss, as an example, that is something that almost every insurance company is most concerned about. You know, because the earlier you get notified, the faster, better they can make a decision. And this is a huge need. So you think of any insurer, any employer, wherever these losses occur, that's a big part. Or if you take in education, for example, if you have a tutoring use case, you know, that's something that's important to almost everyone. I think what's been for us, it's been about finding those use cases that can create, you know, great opportunities to scale, right? And so I think that's been one of the things that we have also seen. So I think that those, but there are also a lot of other criteria in terms of how you determine the use case here. Like I said, this is where you wanna work with people who know that industry. You know, we never knew. I mean, when I started, I didn't know the industry of insurance really well. I still don't think I know it as well as most of them do, but yeah. ... And you talked about another 12 verticals in total. Is, I mean, once again, is there a sort of limit to where you can deploy this technology? Is, you know, twelve the upper limit, or is it a question of just focusing, narrowing the focus to- No, I think there is, I think there is significant room to grow in other verticals. You know, for us, you know, when we went for the... You know, because most expensive to build the models and the use cases for the first vertical, and then it's been progressively getting easier and easier. It was still expensive. Mm-hmm. The second one is not as expensive, but still. Today, the cost of entering a twelfth vertical is not as expensive as the cost of entering the second vertical. I think from our perspective, we'll continue to do that. I think, you know, I just want to touch back on one other thing here. One of the other things about this industry vertical that you're talking about is this becomes a revenue generator for many of these end customers or a significant revenue saver. It creates really high-quality paying customers as well when you go through something like this. That makes a lot of sense. I've got a number of questions. I just wanna make sure whether anyone would like to ask any questions from the floor, just in case, but otherwise, I'll carry on. I mean, one of sort of our philosophies on the topic of sort of AI, extractive or generative, has been that, you know, over time, you know, a quality or relevant database will always trump a good model, a good algorithm. I'm interested with, you know, what you've got these models that you're deploying as part of these as these apps, but I presume that they're ultimately based on your customers' data. How do you make sure that that's secure and that you're not gonna be, you know, sort of compromising, you know, what for them must be very important? Absolutely, Tom. That's what I talked about, the sovereign cloud or the on-prem model I talked about here. That's really core to it. It is precisely to protect their data, right? So every company is possessive. At the end of the day, it is their institutional knowledge and data that is going into this application, so they are gonna be very, very protective of that. Which is why we built our own, these specific narrow intelligence models, or people, you know, talked about size, but really specific intelligence models, I think is one way to think about this, that are designed to fit, to make on-prem, to make that unit economics work, and then build these various use cases on top of that. So, customers are very, very protective of it. You know, the way we think, we look at it, our models are out of the box. They come with a baseline set of data for that industry vertical that gets deployed, and then the customer data is used to then fine-tune those models and build it. All that stays within the customer. I think that's really, really critical to, to understand. I think, which is why, like I said, back in 2018, those days, we saw the need for this when everyone's talking about cloud, cloud, cloud. And I think there's a lot of room there for a lot of application, but there's a whole target on mission criticality, where it becomes a challenge. But that's it, you know, all our implementations are on the cloud. You know, we have hyperscalers for, you know, all the major things, whether it is, Google, AWS, Azure. We have this, and we think they have a strong role to play. You know, there's a lot of the rise in sovereign cloud, you're seeing a lot more, all really getting to the heart of this. Really, it's the key to adoption for us. What we've seen is, one, addressing key risk concerns, data sovereignty, unit economics is critical. But then beyond that, like I said, using this low-code AI canvas and ability to build use cases at a very attractive price point. You can build, you know, like I said, in weeks to months, you can build these applications or... And then once you build, if there are certain use cases that are not working as well as one anticipates, they can cut them out, and the ones that are working, they can scale up their ROI, right? And I think that's really the core of what we're trying to do. Yeah. One of the themes from the conference, I think more broadly, is that there's, you know, a lot of investment going into AI, but there's obviously within that, quite a lot of exploration. There's not necessarily a, you know, full confidence on what the precise model that we're working towards. And in that context, based on how you describe it, I can see your offering is clearly quite attractive because you get quite, you know, with relatively low cost in the big scheme of things, you get quite sort of quick results, I suppose. I suppose the question is, are you worried at all about this, that there being sort of bubble, you know, characteristics in the broader market? You know, what, I mean, maybe not a number, what percentage of the time do you end up sort of going in, sort of looking for a use case and then not really proof of concept and having to come back out again? Right. But, Tom, I think this, like I said, I think the way trying to summarize how we are performing this work in the AI landscape versus others, I would say we were, from the beginning, use case first. Mm-hmm. Others were, from the beginning, product first. Mm-hmm. So they were part of looking for solutions. Yeah. We had strong use cases as a starting point from where we went, and I think that's really a critical part to all this. So that's why, like I said, when you're going into vertical, you know, you just went with the use case. I mean, when everyone's talking about the AI bubble, the AI bubble came now, but we've been around for a long time, and we've been, you know, our capital efficient always been very low, and we've been working very hard to get to where we are, right? So when you have to be capital efficient, one of the key things is you have to be right. So you've been very judicious about how you pick and choose use cases. That's really why you go work with these value-added resellers who bring this expertise. So the moment you suggest a use case, they can tell you if this got any legs or not. Whereas if we had raised, you know, hundreds of millions of dollars, the pressure is to deploy dollars somewhere that's worth chasing use cases, and I think that's been the thing for us. I feel like AI is a tectonic shift in what's going on. I mean, you think about this, every enterprise has hundreds of thousands of use cases, and what they have today in their premises is a few hundreds of applications. So the traditional software application, though they were expensive to build and support. But I think there's an opportunity here with AI for a lot of those use cases to become AI applications at a lower price point. So just think about the scale of what's out there, right? I think the opportunities are tremendous. You're seeing a lot of investments in infrastructure that's going on, the compute infrastructure. You know, we're moving away from a network-based infrastructure to a computing infrastructure. A lot of investment going in, and I think there's a huge need for a platform like ours that could make efficient and effective and good use of that infrastructure, and not throwing that infrastructure away on use cases that don't make any sense. That's... I mean, one final question on that, and then we'll, we're vaguely running out of time. But, you know, given your model is, you know, very use-case specific, it's relatively quick and light and cheap to deploy in the context of some of these AI investments. Yeah, but if it really starts working, what's the risk of disintermediation that your customer says, "Well, actually, this is a clearly defined,"- Right ... and successful, you know, process. I could do this without necessarily using a third-party provider? Right. We think about that a lot, right? And so that's what all goes into how we build our platform. We build an out-of-the-box platform where many use cases can be built, that we're not a point solution. If you're a point solution, you have that risk of being disintermediated. But within our platform, there are use cases that are critical, that are price insensitive, and on the other side, you have things, an in that pool and others that are not. Once they're all on the same platform, I think you have a great opportunity here, you know, to be very sticky and makes it much harder for people to disrupt you. Even if there's one use case that you're disrupted, you have the protection of all the other use cases on your platform. And so I think that's really critical why we built a platform that can support many, many use cases. But you're absolutely right. If we were building a point solution, I would be up at night every day wondering who's gonna come and disrupt me. Okay. We're running short on time. It's obviously been an eventful few months anyway, and as you mentioned at the beginning of the conversation, the last few days have been particularly eventful. No doubt you've got a lot of short-term work addressing the investigation, which you mentioned. But if we try and lift our eyes to the horizon, what are the milestones that you think investors should be looking at over the next twelve months to show that you're progressing in the right direction? Right. So I think for us, our whole journey as a company has been about transition. You know, we went from a transition in terms of relying on one partner, and then it's going away, away from that in many ways. But I think, Tom, you know, we saw a few things, like I said before, a lot of people, like I said, that the understanding that the future of AI was gonna be on-prem or sovereign cloud. The need to pay for data, the need for data sovereignty, the need to build use cases at a low-cost infrastructure. Now, the thing with that is, like I said, one of the things we haven't done well is make ourselves known and participate in the thought leadership. But that's something you're gonna hear us do a lot more, for sure. Get us known there. I think we are, you know, we are working very closely with a lot of hyperscalers, so I think we're gonna start to really make that a top part of our focus. I think we have a huge opportunity marketplace, which we think we have the best-in-class product. I think we have a very powerful go-to-market strategy, and the area that we really want to get past here is this, making sure that other people are talking about the capabilities of our platform. I think it's gonna be critical. I think so that's, I think, my number one lesson out of all this. The challenges of going through this process of a de-SPAC was not about a de-SPAC structure. I think we became very vulnerable because, you know, we saw things out there before others, and we just weren't participating in that thought leadership. If we had talked more about this and made people understand why these things are important, I think, you know, they would get it, and I think that's the one takeaway I have. But I think that's gonna be very important for us. You know, a lot of people have invested in this company, and they've, you know, in the last few months, and they're lost their money, and that's something that's for me heartbreaking with all these things. And so my goal here is to really focus on, you know, seeing what I do to get, making sure that we continue to deliver, you know, a great business and value to them. You know, like I said, we have a best-in-class product, a great strategy, and I think this is an industry where almost every technology partner, company, software company, they all need an AI asset like ours. So I think it's... that's really my goal, and we're gonna really work towards getting this turned around and go forward. Listen, thank you very much for spending the time. It must be a very busy period. Yeah. Thank you, everyone, for attending, and look forward to seeing you soon. Thank you. Thank you, everyone. Thanks, Harish. It's very nice to meet you. You know, we actually got connected back by Alex a few months back. And he and I used to work together a long time ago, but