 From theCUBE Studios in Palo Alto in Boston, connecting with thought leaders all around the world, this is a CUBE Conversation. Hi, I'm Stu Miniman and welcome to a CUBE Conversation. Always love when we can dig into practitioner discussions in one of the editorial themes I've been really looking into in 2020 has been discussion of serverless. So really excited to welcome to the program, Dave Anderson, he is director of technology at Liberty Mutual, coming to me from Ireland. Dave, thanks so much for joining us. Thank you, Stu, delighted to be here. All right, so I think most of our audience is probably familiar with Liberty Mutual. You work in the software group, Liberty IT is part of Liberty Mutual. If you could just start us off, give us a little bit about your background and your group's role inside the organization. Sure, Liberty IT, or something 20 years ago, is really as a sort of internal software house, part of Liberty Mutual Group, or part of Liberty Mutual Technology. So you kind of work across all the markets in Liberty Mutual and all the kind of global locations. My role as kind of director technology is really think about what's the technical direction of Liberty IT. I can lead the architects with my group. I'm really thinking about a global architecture of Liberty Mutual and how can we provide business value in the Liberty Mutual going forward. Excellent, so Dave, I guess, what is the overall business and IT relationship? When I think of companies like yours, usually their M&A comes in, there are growth expansions and digital transformation's been one of those buzzword discussions, but absolutely you need to be close to your customer. There's lots of services that you need to provide online. How are some of those overall dynamics impacting how IT is supporting the business? That's a great question. I mean, technology has always been the key differentiator for Liberty Mutual, even as my group was set up, like I said, 20 years ago, it was always seen as a differentiator. It's something that we need to kind of be very good at. We've always been quite close to be in the cutting edge of technology. Any companies would say, you know, we're not an insurance company, we're a technology company that sells insurance. We are an insurance company, and kind of very, you know, that's very important, but we also need to understand how using the latest technology, i.e. the cloud providers, really helps us deliver value to our business partners and customers. So it's critical that we have a very tight partnership with our business partner. Excellent, so yeah, 20 years, you know, a lot has changed in that time. Give us a little bit if you can, you know, share a snapshot of where you are, kind of in the cloud discussion, and you know, what are the relationships between kind of the infrastructure side of the team and the development side of the team, and I'm expecting that to lead us towards the serverless discussion. Sure, I mean, I joined Liberty about 12 years ago, 13 years ago, and I actually started in a lot of the digital channels, designing systems with digital channels, and probably almost 10 years ago, our CIO, James McLennan, quite visionary, started kind of pressing the public cloud agenda, and sort of discussing public cloud as a potential future. It was a really exciting time, because I think the infrastructure development, we all started to kind of discuss what's the possibilities of public cloud, and as you know, the cloud itself, it probably changes every year, it's redefined, there's new capabilities. I'm not sure we could have envisaged where we are today back when we started that conversation. Like every large enterprise, the initial conversations around, you know, how do we enable this, how do we make this safe, how do we protect our data, all the usual kind of questions you would have, but you know, we kind of really joined together at the various different departments, we thought, well, how do we move kind of the enterprise forward? And as well, I mean, we're all like a global capability for cloud, it was very important, and it's bringing up kind of velocity, that we can deliver value quickly to our business partners. So we kind of, we didn't do it for technology's sake, we did it to find real value for the business. And one of the really interesting things that we talked about is we shied away from counting how many virtual machines do we have in the cloud. That wasn't really a good metric for us, which they will, how are we delivering values to our business partners? That was kind of the metric that we chase and continue to chase. Well, that's excellent, how you kind of laid that out. I'm wondering if you could help extend that and bring us into, you know, where serverless fit into that discussion. I loved how you say, you know, it's not about, you know, number of VMs or, you know, the new shiny thing. So, you know, what was it that led to your first use of serverless and, you know, bring us a little bit along that journey that you've been going through? Sure, well, one of the things that I've always found critical working in technology is that curiosity, that kind of search for what's next. So within my group, I always charge my people to say, like, you know, where do we need to get to? What is three to five years out? And we've a lot of really fantastic years right across Liberty Mutual, quite open-minded. They, you know, dig into the thing ahead. So one of my team was, I think, was re-invent in 2013 where they launched Lambda in 2014 and he came back really, really excited. They can build their first small application. It was actually a document generation. I think they were using some proprietary system. So they built a document generation solution and couldn't believe the amount of, you know, run cost that was saved. It probably knocked something like 97% off the cost. Couldn't believe it. And so we started to think, wow, this is potential. But back then, you know, again, five to six years ago, the stack was very immature. There was a lot of things you needed to figure out, like your observability, the developer environment, you know, there's a whole bunch of stuff that wasn't quite right. So it's something that we share through our peers across the organization. And we talked about it. We, you know, we really started to kind of think, well, this is interesting. This idea of serverless or a managed service, it started to really change how we thought. And it really started to make the concept of a cloud native application, very real. Because we started to think about cloud native architecture, cloud native application architecture and that started to really flip how we thought. So it's just been a real journey those past couple of years. And the big thing for me is we started with engineers thinking of cloud as a mindset that necessarily is a platform. That opened the door to a lot of possibilities. Yeah, it's really interesting that you said that, you know, often time we say, you know, cloud is an operational model, you know, not a physical location. Are you using multiple clouds today? Yes, we probably, we're trying to have a multi-cloud strategy. And really, I mean, the very clear is serverless for us, it's not just function as a service. It's not like we're just using kind of something like Lambda. It's really about that idea of using managed service and thinking about evolution or architecture. But how can we kind of, you know, try and cut out anything that is effectively not differentiating? There's a great term which I always like is the stack policy. It's sometimes as technology companies, we get obsessed by the stack and we think that the piece near the customer is quite easy. I think from a technology perspective, we need to think about, we can deliver the most value by making the customer experience kind of best. On the end, can it be even rent a stack from whoever meets that need? Yeah, no, it's really, it's a fascinating discussion. I've seen even today, it's, you know, well what is, you know, you say serverless functions as a service, a lot of it is I don't want the developer to have to think about those underlying layers. Which reminds me of the discussion we've been having about platform as a service for more than a decade, but has was supposed to be, you know, platform independent so I could have my code wherever it goes, serverless today, you know, right now, I mean, you talked about Lambda, Amazon, there's certain things that I can only do on Amazon. There are discussions and working groups and the cloud native computing foundation working at how we can do, you know, serverless functions as a service that could span between multiple clouds or use the same sort of code. So how do you look at that space? You know, you talk about, you know, cloud native, how do you make sure that you're leveraging the technologies of a specific cloud but, you know, I guess I'll throw out not being locked into, you know, any one provider. I think about it for me, it starts with the empowerment of the engineering team. We talk about a serverless first strategy. That means that you've got the capability to build anything you need to, but you'll rent where you can. We had a fascinating story, one of our test stories is a company called, or a area called WorkGrid, WorkGrid Software is one of the companies that we spun out of Liberty Mutual. That was a project that we had an internal digital assistant that we built with some of our teams be back four or five years ago. And our CEO, Gene as a Glennon decided to fund that as a kind of a startup. So they broke off around three years ago and that initial team had four people and that engineering team. So they kind of decided that they would be serverless first in their approach. So they didn't have time to think about operability or right for portability. They had a realized business value really quickly. So they took an evolutionary architecture approach which meant they kept incrementing and iterating and delivering value where possible. Like what's the next best thing they can build, deliver value for the customer. So when you think in that regard, if you ever come to an Amazon of a great, talk about no one way doors, keep a two way door, don't lock yourself in the anything, make decisions that you can always build upon. So with that kind of constant iterative work of our teams and that serverless first strategy, that means that when you do rent a service, if you need to change to another service, it's just a matter of, if you've got your boundaries set up correctly, it's fairly easy to get rid of that. You dig yourself in deep to something and then that's the difference. So I think there's an engineering mindset and culture that we certainly have brought with in our teams that they kind of go fast, focus on business value and kind of try and be sort of cloud native in their outlook. Excellent, yeah, that I just heard Andy Jassy in the AWS Summit online, you know, talking about those one way or two way doors. So sounds like from your standpoint, serverless is a two way door for architecting in your mindset. Absolutely, I mean, I think it's really, for me, it brings architecture back into the team. It's one of the really nice things is if your team are kind of using managed services, that focuses on business value. If our infrastructure itself support that type of team, then you have minimal handoffs within the team. If you have a single team, it's their job to create value, engineer the solution, make sure the security is good enough through the operations, the visibility. It's all contained within one team. We get a huge responsibility within that one team. As an engineer, that is super powerful and super huge autonomy. So we can talk with the serverless engineer. And for me, it's been absolutely fascinating to see teams come into this environment. And once they understand, you know, that event-driven way of creating their systems. And I use the word systems drawing applications to create the event-driven systems that constantly in the building upon. It's just fascinating to see where they go. It's, you start to see the creativity and innovation from a lot of engineers. So it's truly unbelievable to watch. It's really quite cool. Dave, I'm curious, when you look at the application portfolio that your team manages, how do you decide what goes serverless? Is it, you know, new development? All goes serverless? If they've been a migration, you know, how do you look at the overall application portfolio that you have? Sure, well, it kind of depends, I suppose, on, I mean, I'm not going to sit here and say that we're going to refactor everything the serverless. You know, I think when you do migration, you have your, you know, your six or seven paths that you can go down and you do what's best for the business. But for new development, it's definitely interesting. We haven't found many use cases that are really a bad fit. It's a spectrum. We may decide, you know, what different servers do you use? We built a system last year, which was absolutely fascinating to see. It's like a financial kind of aggregation system where we do a lot of our accounting. So it was kind of called serverless ETL. We're trying to run like a end of month batch system. You take a lot of accounts from different countries and kind of pipe them into a kind of general ledger. Not something I would have thought about for serverless, to be honest, but then when you think about it, and some of our kind of engineering needs, as they put this together, they kind of designed this fantastic system using serverless workflows, because you're taking lots of various different types of data orchestrating them in this single destination. And we've went live with that this year. And I think one of the month ends that they recently ran, I think they ran something like 100 million transactions, costing relatively low cost. And of course, being a month end system, the rest of the months, there's zero cost. You don't pay for idle. We only actually pay while it's running in that month end process, or maybe, you know, a few hours or something. Yeah, Dave, you talked about in the early days, 2014, when Lambda was announced, re-invent, when you first started using in the first year or so. There's the maturity of that ecosystem solution set. Where do you see things now? What's working well? What's on your wish list to kind of mature or increase overall functionality to help? I think the developer environment and developer experience is a big part. One of the key messages in trying to kind of get into the NDR culture is code is a liability. It's not NASA. If you have a bunch of engineers writing lots of code, that effectively is a liability. There's no business asset in that. The asset is in the system that you create. They're trying to get engineers into the mindset to write less code than they actually engineer systems. So one of the things that we've been trying to do is maybe using patterns as building blocks. People, they kind of like a Lego building block way of creating their systems. They're using some things like CDK, for development kit patterns, using the well architected process to make sure that teams are looking after their costs, their security, their performance, their reliability, and their kind of optimizations. So some of those things are really important that aims that whole ownership of an operational view of their systems. And also even things like observability. When you create a system with a lot of advanced plan around, it starts getting complex. But then if you do that correctly, you can start to layer on what data insights can be built on top of the system. So it's really opening up aims a different way of working. And then of course, there's lots of operational challenges when you get in the more complex environment. So as we often say, it's not easier, different, build difficult to build these systems, but it's different. So it's definitely easier, but better. All right. How about anything that you're looking out towards the future? You talked about the early days when you look at these, and while you're not necessarily throwing the light of shiny thing into production, there's that curiosity. So what's exciting you now? Anything else looking forward that you could bear? I think, I mean, one of the fantastic success stories we've had is with a project we've called Virtual Assistant. And really that's your question. It's about how teams can properly work at MVP. So one of the things that we really find fascinating is when you put a good engineering team, and I mean a team who are really solid engineers, we then layer on the cloud and secure the best practices and certification. We then put them in a supported service environment with a real business problem. They then create an MVP. So our Virtual Assistant team created an MVP where they would start to see if you call them to the call center and you've a fairly straightforward request. Like if you have an auto claim, you might say, well, when can I pick up my rental? If you've already done your claim. We have an NLT, a natural language assistant that can help you out with that conversation. So when you start the MVP systems like that, you can start them off with a fairly small traffic until you actually tune that until it's going to perfect. And then you gradually scale up and add on other data-driven potential AI services, you know, integrations to that. So I think when you start to take that MVP approach and have a very, very narrow solution, see what that's like in the wild and then start to scale that up. I mean, we scale that system from taking maybe 30 calls, you may be taking a quarter of a million calls. So it's fantastic how you can start to scale these systems up. So I think what I'm really looking for is more kind of support to see how we can, you know, it's the art of the possible. How can we use this tool set and this service mindset to create really fascinating, you know, business applications? Because when you get into that kind of creative conversation with business partners, I mean, they don't want to hear about Lambda or events or observability. They want to say, well, what can I do with AI? What can I do with voice? What can I do with vision? You know, you start to open up really fantastic, you know, conversations like that. So I think that's, I'd like to see more of that, that kind of creative product development. Excellent. Well, yeah, Dave, so important that you brought back as to how, you know, I've seen the business working together. It's not about, you know, the technical, you know, widgets and knobs or anything, but you know, the services and the value that ultimately you can provide to the business and, you know, the impact that has on your ultimate end customer. All right, Dave, thanks so much. Real pleasure having a chat with you. Thanks, Drew. I really appreciate it. Thank you. All right, be sure to check out thecube.net, lots of backgrounds. If you go hit the search, you can actually type serverless, find out more about what we've been covering as well as what events we will get in the future. I'm Stu Miniman, thank you for watching The Cube.