AI Readiness Ep 18: Crossing the AI Chasm with Jack Levy at Allvue

So, we've got Jack Levy here with us today. He's the head of Data Strategy at all of you. Jack, thanks so much for the time, mate. Pleasure. So, for those that have been following along, this is a continuation of our AI readiness series that we've been doing over the last couple of months. And yeah, very much excited to get Jack's perspective on this. So, I'm sure it's a natural place to start is, you know, what have you been doing, Jack? You know, how have you been getting ready for AI and, you know, what's been your approach so far? Yeah, great question. Thanks again for having us on. I'm about five months into my journey here at Allview, where we are, luckily and fortunate to be in a position where sort of people that have been here before me have been thinking about this for a very long time. What that means is we are really well positioned at the moment to create value for our customers in a meaningful but also responsible way by making use of some of these new technologies. Specifically, I would say that some of my colleagues had a foresight maybe about five years ago to understand that this all starts with data. You know, you can't have a conversation on this topic without hearing the cliché about garbage in garbage out. And I think they've realized that very early on and invested in building out a data platform that ultimately does a lot of the heavy lifting to cleanse and organize our customers' data into a form like that is AI ready. We launched our first AI products, our Andy agents earlier this year, and we're really only able to move back quickly because of, you know, about that good work that was done before. Look, I think that more broadly as I kind of look across the industry, let's be honest, I think chat GBT really started Gold Rush and kind of democratized AI for both the kind of non-technical employees, but also for consumers, which is great. But I also think the flip side of that is we can all think of examples of applications where, you know, maybe rushed out AI features led to a less than optimal customer experience. You know, I can think of chatbots that get stuck in infinite loops and the like. So, yeah, look, we're in early days. But I would say that we're starting on that inflection point of maturity in terms of the approach to AI. So, look, we can get that immediate value. For me, I think it's things like prototyping or doing a POC. I can spin up a UI in a matter of minutes with some decent engineering, put it in front of the customer and get that immediate feedback. But, you know, coming back to my fundamental point, any sensible long-term AI strategy requires you to get the data in all the first. And I think that's true amongst the customers I talk to and peers of one in the industry, you know, that penny has dropped. If you really want to make the most of AI, your dose of foundations need to be absolutely solid. You know, on the people front, if you spend time on LinkedIn as an idea, you probably can't read more than three scrolls down without seeing something around. All product managers are all developers and we redundant because of AI. Look, I fundamentally disagree with this. Obviously, it will take away parts of the job, but I like to think it's the mind-numbing part. I've been a product manager or variance thereof for the last 15 years. And I hate updating JIRA stories or writing up notes from customer calls. And that's part of the job that has largely gone away. And I think that's a great thing because it frees you up to talk to customers and focus on what you should be doing, which is solving problems and delivering value for customers. On that specific point of product managers, though, I think a lot of people I talk to in the industry are really asking that they do need to step up their data skills. And more broadly, the old-fashioned way of product technology and data being silos just doesn't work anymore. I'm seeing once customers and again peers in the industry a huge move towards setting up cross-functional data technology and product teams. And I think this is great because it creates a huge opportunity for talented products and data professionals in the industry. Really interesting point. Yeah, I think just to go back to your first point about having the data estate ready, I mean, that's something that I don't necessarily think has changed in the industry over the last sort of five to 10 years. That's been something that people have been on a journey from a transformation perspective for a while, haven't they? I just think that the kind of allure of AI has meant that people are like, "Right, actually, do I have to start taking this seriously now?" I think it's been my kind of perspective on it. But yeah, I think that one of the motivators for doing this series was that data professionals in general have been put under a significant amount of pressure in recent years because it's like we need results now, we need to get AI in place, we need to get these things out. And that's almost been a new phenomenon as well, hasn't it? So I'm curious to get your perspective on that a little bit about some of the pressures from a product perspective for people to all of a sudden be wanting to be everything now, because it's probably always been the case, but have you felt that kind of shift in pressure? Absolutely. I can think of the times in my past career where we've had that conversation around where are we going to invest? And I think sometimes before this, what I've referred to as kind of a gold rush, it was sometimes difficult, particularly in siloed data organisations to make the point for investment. So why do we care about data governance? Why do we care about cleansing? Why do we care about data mastery? It was almost hard to kind of say, "How does this move the needle?" And in a lot of cases, a lot of businesses, that may be deprioritised while we're focused on features or stuff that deliver more short-term revenue. I think the seismic shift here is that you can't unlock the new revenue that AI potentially offers unless you do this work. So, call it a carrot, call it a stick, but this isn't optional anymore. And I think that is a good thing, not just because it unlocks AI, but actually forcing people to take data a lot more seriously than maybe they previously had to is hugely important. I've been in and out of product for 15 odd years, and I think the amount of times I've had conversations with the customers about not being able to report because data is not in the way that they thought it was, or it wasn't being collected in the way that they thought it was, I think this is almost a very big stick and also a very big carrot for getting that stuff right and where you haven't, making that investment to go back and either fix or update or modernise the data stack. And it's a lot easier to make that case because the payoff is a lot more clear. Now that AI offers new, very clear revenue opportunities, both in terms of product features, but also in terms of what it unlocks, there is more emphasis and effort to get this right. Fundamentally, data quality, data governance is no longer optional and it's absolutely table stakes for businesses if they really want to take advantage of all the benefits that AI offers. Yeah, no, I think you're right. And just to go into the second point, sorry, about what you said earlier with regards to previously having those silent teams, that feeds into another point of interest for me around creating the right environment. Like you mentioned again, you scrolled on LinkedIn, fame is happening everywhere. Is that an example where you've seen those different teams within a product come together and culturally work together and see the benefits of AI and actually want to embrace it and have the transparency around where it's going to be able to add value instead of it being like, right, we've got a secret project over here happening on AI and the rest of the workforce is worried about it. It can be, in my experience, very counterproductive towards actually embedding it because teams, they don't want to collaborate. Then all that does is slow down the transformation process. Yeah, no, absolutely. And I think when I think about this ultimately, it comes back to the first principles, right? We're using to solve customer problems and hopefully solve them more quickly. But if I think about my current organisation, we do have a dedicated AI and data teams, but we're very much embedded with the core business. We have subject matter experts in private equity and private credit. And there is no point in us building anything unless we're talking to them and talking to their customers too. Everything we do is to deliver value for our customers and ultimately build out the capabilities we have. So I think that cross-functional approach, that kind of open door policy and falsely breaking down the silos, which luckily both our board and our executive team are very much embracing and supporting. I think also it does start with education. There are some people that come in with huge expertise of the technologies behind this, but maybe less of the subject matter expertise of the end customer. So I think working on both is important. I think we're very lucky here at Allview to have strong expertise in our leadership team. We've done a fantastic job of helping colleagues across the business to understand both the key concepts, but also the opportunities associated with AI. The sector is moving faster. So I think it is increasingly becoming everyone's job to stay on top of their own knowledge and keep learning. This is moving so quickly, and new developments are coming on onto the market weekly. So this isn't one and done. I would say one resource that I found very helpful over the last year is Richard Susskind's How to Think About AI. So anyone who's struggling or feeling lost, I would very highly recommend giving that a read if you want to go from zero to one quite quickly. So I spoke a little bit to the opportunity and the education piece, but the flip side of that is guardrails are hugely important. I think with all great horror stories of employee has popped some IP sensitive data into the free version of GPT, and it's gone into hands where it shouldn't have done. So I think there are some fundamental things like making sure you have the right commercial licenses in place if you are using these tools and you have controls and policies internally so that people know what they can and can't do. Here, we've actually created our own GPT-style front-end that de-rescues some of that and makes sure that we're only pointing stuff to supported models. But yeah, I would say it's a combination of education, but also putting that governance and those controls in place and making sure they evolve. So more than any other area I think of, this is fluid and it's evolving. So that has to be a continual process of learning education and updating to make sure that you are taking advantage of this, but doing it in a responsible and mindful way. Yeah, I agree. Yeah, it's a bright... Well, not surprising, but also speaking to government people. We've had a few governance people on this podcast. I don't think they were then so popular over the last 12-18 months as they've been before, but it just speaks to the way that you have to be managing these tools and they are powerful in a very responsible manner. Absolutely. Especially if you're dealing with sensitive data and financial services data, it's absolutely must and there are some great tools out there that can help you do that. I think for us, we are ultimately custodians of our customers' data and that has to be paramount in our thoughts at all times. So around policies, making sure that we have attorneys who are skilled in this to make sure that we stay the right side of those agreements is, again, table stakes. It's not a ball we can afford to drop. As I say, harness the opportunities, but be responsible and be mindful about the data you're using is hugely important. Awesome. Any of the thoughts on creating the right environment? Just wanting to touch on that before we do get into some of the stuff around different use cases, but is there anything else from, again, creating that right environment where we touched on governance just now? Is there anything else that you've seen that gets people comfortable with this and actually driving it forward? Yeah. I think we've done a lot of these hackathons to try and bring people in and prove out the value of this. I think in the time I've been here, which is about five months, we've done three of these, and it's a great opportunity because, again, I talk to creating cross-functional teams, and that's a great way of doing it. You can bring in engineers who understand, you can bring in the subject matter experts, bring in product people, and you're solving a very short-term problem, but it's a great way to learn on the job and see almost immediately the benefit of what you're doing. So that's been a huge enabler for us to really grasp this opportunity. But yeah, more broadly, it's education, creating that space. I think we've set up Slack channels, for example, where people share new developments, news, interesting articles, and really making it part of business as usual. It is business as usual, and that's really the mindset shift that I strongly recommend that people start to take. Awesome. Yeah, so again, into the use cases and stuff, people are really interested in it. I suppose it's natural, isn't it? People want to see what the kind of output is going to be of embracing AI. So obviously, it's much as you can't be interested to hear some of the compelling use cases, and I suppose more generally as well, just it does feel like we're getting to the stages where people are demanding to see an ROI on these efforts, but we're getting to that later. But yeah, what are you seeing that's particularly been compelling? Yeah, I can certainly speak to this. I think we're building out a suite of new products at the moment. So the first opportunity we're seeing, and I mentioned earlier, is in that kind of prototyping or proof of concept work, what would have usually been work for a UX team to kind of go in, figure out and build prototypes that could take a week or two. We can do that in minutes. So we can almost sit with the customer now, talk about an idea and with some decent prompt engineering, we can start our UI, start talking through it, change it on the fly. So that whole requirement gathering process is massively changing. It's hugely more efficient. And I think the scope for going wrong is significantly reduced by leveraging that. So for a personal level, that's been a huge enabler to us as we start to build out new products and services. But you know, on the flip side there, as you correctly referenced, my title is had a big strategy. So a lot of my focus has been on how we can use these technologies responsibly again, to help us with organizing, maintaining the data. So one area where I am personally seeing gains is data normalization. What maybe five years ago, it would have taken engineers many, many months to write mapping scripts to kind of take disparate data and put it together is now, you know, could be done. I won't say minutes, but it's gone from months to days. So, you know, that quantum leap is huge. Again, kind of data mastering. This used to be something that you would go to a specialist provided to do aspects of that job in terms of things like entity matching and de-juprification. Again, that's, you know, that can be done, or at least maybe the 80% of that can be done in a much more compressed timeframe. So there's huge opportunity for us both in terms of the way that we interact with customers, and we do all kind of requirements gathering and really focus on the problems we need to solve, but also from a more practical level about how we kind of manage data and sort of harness internal efficiencies to get that data ultimately in a place where we can start to build these fantastic AI based products. Awesome. Yeah. You know, how much, I mean, I guess as a product person that's a huge driver, is it products and just the use change effectively the same thing? Or is it, you know, how are you kind of, I guess, my own curiosity? Like, how is that differentiated? Is it kind of baked into a product? Like, you know, obviously, how is that different than someone who's in more of like, I suppose, a data management role? I think the line between the two is blurring, to be honest. I've had a meet up just a couple of weeks ago, and that was very much the type of conversation. Like, how much longer are we actually going to talk about data owners versus product owners? Are they not converging or being the same thing? You know, my role is very much a hybrid. I come from more of a product background, but, you know, over the last five years or so, I've kind of veered as much more into data. But, you know, ultimately, I don't see this job as being fundamentally different. Everything starts with data, let's be honest. And I think, you know, if you're a good product person, you're looking at data. If you're a good data person, you should be using kind of product management and talking to customers and those sort of techniques to ensure that ultimately the data you're taking and you're managing are responsible for is data that's valid, data that's important, but data that's ultimately organized and managed in a way that delivers the best possible value to your customers. Awesome. And closing thoughts on, you know, just your advice for anyone. It does sound like, you know, you're much further ahead than many of the other people, actually. We've, you know, we've had on this podcast. So what would your kind of advice be, you know, elevated pitch style, if you like, to not support you on the spot? But to those people that are, right, the kind of influences of their journey and want to make sure that they make, you know, the first few steps, the right ones, I suppose? Yeah, I think it's a great question. I think the first thing to say is, you know, there is a continuum here, right? So we are definitely on the journey. But, you know, I speak to people every day who are far more knowledgeable about this and far more focused on their journey than I am. You know, ultimately, I say, come back to first principles. What is it a problem you focus relentlessly on the problem you're trying to solve? And, you know, don't, don't lose sight of that. It's very, very easy to think, you know, oh, I'm going to plug in a bot or I'm going to do something with an agent because it looks good and excites customers and investors. But ultimately, that's just a solution looking for a problem. So don't lose sight of the problem you're trying to solve. And look, if, if an AI based solution is the right solution to that problem, go for it. But it won't always be the right solution. So, you know, be responsible, be mindful. And I think, you know, the second thing is keep learning, you know, we are blessed that we're in a world where there are so many resources out. And I think the third thing is like, get your hands on as well, you know, start building, start experimenting, you know, as somebody who has spent most of his career building enterprise grade software, the idea of vibe coded software going out to production sometimes scares me a little bit. But in terms of proving out a concept or, you know, just educating yourself, there's never been a better time to do this. So that would be my my kind of three main points to to get yourself on this on this journey. Awesome. Jack, thanks so much for your time. That was really, really insightful. Let's do it again soon. Cheers, mate. Thanks, speaker. Thanks.