2876: Are We Prepared for the Evolving Landscape Where AI meets DevOps?

[00:00:00] Is the advent of AI in DevOps signalling a new era, or simply the next phase in an ongoing evolution?

[00:00:10] Well today we're going to delve into this intriguing intersection with Elizabeth Lawler, CEO of AppMap.

[00:00:18] She's a visionary in the realm of software development and AI, and in an industry where change is the only constant.

[00:00:27] I want to learn more about how the fusion of AI and DevOps is redefining the landscape, prompting a re-evaluation of workflow, skills and ethical considerations.

[00:00:36] And Elizabeth is going to be bringing a wealth of experience and insights into how DevOps principles are not only influencing AI implementation, but also being reshaped by it.

[00:00:47] So from the necessity of rapid upskilling to navigate the AI infused landscape, to the crucial role of openness, transparency in overcoming AI's notorious black box problem.

[00:00:59] Today we're going to explore the multifaceted impacts of AI on DevOps workflows and so much more.

[00:01:07] Now before I get today's guest on, quick shout out to the sponsors of Tech Talks Daily.

[00:01:12] Because in today's remote first world, I think settling for outdated managed file transfer solutions means ultimately you're risking your sensitive data.

[00:01:21] But if you upgrade to Kiteworks, the gold standard insecure MFT, boasting FedRAMP moderate authorisation, Kiteworks isn't just secure.

[00:01:30] It's a complete transformation of how your business handles file transfers and communications.

[00:01:36] So say goodbye to compromise and hello to unmatched security and efficiency.

[00:01:41] And you can do that by making the switch to Kiteworks.com.

[00:01:45] Visit Kiteworks.com to begin. That's Kiteworks.com to secure your data and empower your business.

[00:01:52] But now let's get today's guest on.

[00:01:55] So buckle up and hold on tight as I beam your ears all the way to Boston where it's snowing in April believe it or not.

[00:02:04] But despite the cold snap, Elizabeth's going to be uncovering the dynamics of this transformation,

[00:02:09] the emerging opportunities for professionals and the pivotal role of developers in steering the future of technology.

[00:02:18] So a massive warm welcome to the show.

[00:02:21] Can you tell everyone listening a little bit about who you are and what you do?

[00:02:25] Sure. I'm Elizabeth Lauer. I'm the serial startup founder.

[00:02:30] I'm currently the CEO and founder of AppMap.

[00:02:32] AppMap is a no effort developer centric observability platform that works in the Code Editor and CI

[00:02:39] to help developers better understand their complex applications and deliver high quality,

[00:02:44] performant and secure code.

[00:02:46] And we're also the makers of Navi AI, which uses runtime context to help developers understand

[00:02:52] and troubleshoot enterprise applications with AI.

[00:02:55] Navi has got the context of an experienced software architect.

[00:02:58] So it knows the whole application and can provide better recommendations.

[00:03:03] And last product we have is there's so much buzz about AI encoding and software engineers.

[00:03:09] So we're launching an AI software engineer battleground that's open source called Code Challenge AI.

[00:03:15] So people can test the strengths and weaknesses of various tools, context LLMs.

[00:03:20] And Code Challenge AI is an open source, open context repo because we believe in transparency

[00:03:27] and developer and AI tooling.

[00:03:29] Fantastic. It was a pleasure to have you on the podcast.

[00:03:32] I must admit whenever I hear anybody talk passionately about that developer community,

[00:03:38] I almost want to channel my inner Steve Bullmer and just say developers, developers, developers, developers, you know.

[00:03:43] Totally.

[00:03:45] Yeah, it's even more important now, especially here in 2024, the AI workflow is witnessing a transformative shift.

[00:03:53] Almost reminiscent, I would say, of the DevOps revolution.

[00:03:57] So could you share your perspective on how you're seeing AI reshaping software development workflows

[00:04:03] and maybe some of the parallels you might be seeing with those early days of DevOps?

[00:04:08] Definitely. In fact, the language you're using in the industry is so reminiscent of the early days of cloud and DevOps.

[00:04:14] You know, large language models as like a multipurpose common infrastructure that anyone can use to deploy applications faster

[00:04:21] is a lot like the same language we're using in the early days of cloud.

[00:04:24] I still think EC2 is like one of the most important inventions in software development,

[00:04:29] and it's contributed to the achievements we've seen in AI now.

[00:04:33] But with every transformational platform, I think people are rushing in to figure out like what's the bigger strategy?

[00:04:38] How do I apply these tools? And the hype language is a lot the same as it was then, right?

[00:04:43] You'll never need a data center. Well, there will be no more IT jobs.

[00:04:47] All the system engineers will be eliminated. Well, that didn't really happen in the DevOps era either.

[00:04:52] We just ultimately like got a little smarter and better at our jobs.

[00:04:56] We did a little more code as infrastructure, but we didn't ultimately eliminate all those positions.

[00:05:00] So I kind of feel the same, you know, kind of temperament towards the AI hype cycle

[00:05:06] and the narratives now around you won't need marketers, you won't need software engineers.

[00:05:10] AI will do it all. It's fascinating to see how fast Hello World applications can be generated using AI from scratch.

[00:05:18] It really is going to be cheaper and faster to build on.

[00:05:21] But I think after the initial hype, we like kind of the realists show up and they say like,

[00:05:26] how are we going to really translate this? How are we going to get this technology applied to enterprise legacy,

[00:05:32] the data center or worse yet, like even things that never made it to the cloud from the mainframe, right?

[00:05:38] Like, you know, that's where the domain experts come in and they start working on how to really translate this for that enterprise

[00:05:44] and start building the missing pieces to make the stack and solutions better.

[00:05:50] And, you know, how are you going to translate it for real serious workloads on real applications

[00:05:55] where you've got security and compliance and governance controls that you need?

[00:05:59] And, you know, I think we end up ultimately probably where we ended up in DevOps,

[00:06:03] which is we'll build a whole ecosystem of tools to monitor this new infrastructure that we are now leveraging.

[00:06:09] And we will have, you know, tools to monitor this transformation for us because it's the second wave builders are coming now

[00:06:16] and they're really thinking about how to get this tool, these technologies in the hands of the majority of the developers who need them,

[00:06:22] who are, you know, really that that's really where they're at.

[00:06:26] Yeah, I completely agree with you. And as you said, we've seen all this happen before.

[00:06:31] We've been here many times before we saw the arrival of the internet and then the smartphones and the gig economy.

[00:06:39] I would argue that anyone that buys into all the doom and gloom that we're seeing right now is Google jobs that didn't exist 20 years ago.

[00:06:48] And there were no mobile app developers, cloud engineers, social media managers, podcasters.

[00:06:54] The world was very different. Yes, old jobs went, but new jobs came along.

[00:06:57] And I think the key is bringing people along and re-skilling, retraining people so that we don't leave anybody behind.

[00:07:04] But of course, they totally. Yeah.

[00:07:07] And I'm just going to say that the integration of AI into DevOps or some might say the advent of DevOps 2.0 is marked by this significant fusion of principles from both fields.

[00:07:18] So I've got to ask how do you see DevOps principles enhancing AI implementation?

[00:07:24] What would you say are the most crucial aspects of this fusion for AppMapO?

[00:07:29] Yeah, I think it's really interesting because the analysis of the theory of constraints was like a core tenet of the DevOps 1.0 movement, right?

[00:07:38] Everyone was trying to figure out how to get more efficiency, more throughput, and it really fueled tons of the innovation we had.

[00:07:45] And containerization, you know, platform as a service, like all of these different elements came to fruition.

[00:07:53] So when I think about DevOps 2.0, you know, one of the reasons I actually founded AppMap was like thinking about that theory of constraints.

[00:08:02] And one of the recognitions that I had was that if we kind of followed that down, there were some issues that were constraining the pipeline that we hadn't really addressed that I thought that like data and measurement and AI could solve.

[00:08:14] And that was that we'd solve so many problems to delivery and deployment that the remaining hurdle was like upstream all the way back at the beginning.

[00:08:22] It was in the design step, right?

[00:08:23] Where you were trying to develop or design things more quickly, more efficiently, but not sacrificing quality.

[00:08:31] And it was something that weren't really moving the needle on.

[00:08:34] And, you know, you saw this in research that Stripe did and Microsoft did where developers were cycling work, rework on deep technical bugs, performance issues, security issues, and they were calling it toil, right?

[00:08:47] And that's like the ultimate inefficiency.

[00:08:49] Like a poorly designed system upfront is most expensive to fix later in the process or rework or later in the process.

[00:08:56] And that's where we kind of came up with the concept of personal observability or developer centric code observability.

[00:09:01] And that's an AI has now come into that realm too.

[00:09:04] And it's, you know, first we started out with a way of trying to get humans better information first so they could make better upstream decisions and, you know, using those kind of that principle of the theory of constraints.

[00:09:15] And then now I think we're just using it.

[00:09:18] Now we're actually adding AI to that mix.

[00:09:20] And that's helping people make better decisions.

[00:09:22] But like, I think one thing that we learned in DevOps 1.0 that we really need to solve in DevOps 2.0 is that unlimited computing resources that software design basically could suffer under those conditions, meaning that you could explode your cloud bill and you could end up with one of my friends and colleagues call it the DevOps hangover.

[00:09:43] Which is that you can just ship whatever you wanted and you could get it to run in the cloud because you could, you know, you had an elastic resources and you had unlimited compute and unlimited storage.

[00:09:55] And so now we have FinOps and APM mixed bills that are at the board level and cloud costs are spiraling out of control.

[00:10:02] So I think one of the opportunities we have in AI now is to try and fix some of those problems upstream so that we don't propagate them downstream to becoming expensive platform issues, you know, later on.

[00:10:13] So modeling some of these issues and bringing AI to the fore up front in the coding event is going to hopefully help with that.

[00:10:22] And AI software architects are like really good at analyzing complex systems.

[00:10:27] So it's, you know, it is really an efficiency that we can start to bring to bear.

[00:10:31] You know the goal of ultimately running software more efficiently and effectively is like, is the goal we've always been pursuing.

[00:10:36] I think we just have some new technologies to include now.

[00:10:40] 100% with you and talking about new technologies, I think with the evolving ecosystem demanding a new breed of talent, arguably as many human skills as the more technical skills there.

[00:10:54] What skills do you believe are now essential for any professionals aiming to thrive in this AI infused DevOps landscape?

[00:11:01] Because there are a lot of challenges, but equally there's a lot of opportunities for people that have those right skills.

[00:11:06] I was reading before you came on the podcast some of the salaries that are in demand in this space because the demand is outstripping the supply, isn't it?

[00:11:15] Absolutely. I mean, I think even in the future, in the next few years, we might even see a change in the way that applications are written, right?

[00:11:23] We might even end up with more AI favorable programming languages that reduce our use of, say like human reader friendly language frameworks, which we've developed to sort of solve for ourselves.

[00:11:35] So I think there are a few key skills that would be really important.

[00:11:39] You know, the first is critical thinking and code, you know, being able to still read code because you can't put these agents on autopilot yet.

[00:11:47] You know, you're going to need to understand how the systems are built, how the code works, or you're going to have low quality code and subtle bugs that are going to take time to debug and fix.

[00:11:57] And I think developers will need to challenge the output of AI for a while, you know, as they're developing these tools.

[00:12:04] Design and architecture, you know, I think about, you know, there's a whole spectrum of different kinds of, like just understanding what you want to use, like an AI backend, for example.

[00:12:16] There's strengths and weaknesses to different models.

[00:12:18] There's different, there's local models.

[00:12:20] There are centralized hosted models.

[00:12:22] The literature of how you can leverage data with these models is constantly changing.

[00:12:25] You really need to be able to think critically about design and architecture and also be able to keep up to date with all of the different backends and technologies that people are developing.

[00:12:36] You know, things like performance security are going to, I think, remain part of the design decision making that people are going to need to kind of keep their good design as a baseline for software development in this market.

[00:12:51] And I think the last one is really innovation and creativity, right?

[00:12:54] Like AI by nature is trained on historical data.

[00:12:57] So as we leverage these tools in the, you know, to deliver software, to create software, to model systems, like, you know, what you do as a sort of innovator is challenge the thinking of the past, right?

[00:13:10] So I don't think statistical models that are trained on historical data will be able to help in the future necessarily come up with innovative new solutions or innovative ways of leveraging these technologies.

[00:13:20] And I think that understanding, understanding that and coming into helping developers think more innovative or creatively, I think will also help.

[00:13:28] There'll be a poor skill for them in the future.

[00:13:31] And on the flip side of all the excitement and enthusiasm that we've got for this technology, there's also that need to address the black box nature of AI processes.

[00:13:40] Things go in there. Nobody knows what causes the outcome.

[00:13:45] So it's a significant hurdle in software development.

[00:13:48] So what strategies or approaches do you advocate for making those processes more transparent, more understandable and avoiding that black box problem?

[00:13:58] Totally. So there are real deep philosophical movements right now within the space.

[00:14:04] They are absolutely diametrically opposed, you know, and it's more than the sort of open versus closed source that we experienced during the first DevOps movement, right?

[00:14:13] This is AI sitting in the middle of that. But, you know, the ability to inspect prompts to understand their impact or implications, the use of data or IP securely.

[00:14:25] I mean, these were these are really big discussions that are going on in the market and everyone has like at a different place and where they feel they're, you know, their company or their organization fits.

[00:14:36] There are also other discussions going on, like what's the value of really huge LLM models versus small focus models, like some of the work that Nvidia is doing.

[00:14:44] You know, how do we how much emphasis should we place on the models that suck up everybody's data and every data point in the universe versus like building key features and functionalities that can run well on small systems?

[00:14:55] And I think with lightweight local models, and I think, you know, you see these competing models.

[00:14:59] I'm at map is open source and we believe in opens, you know, the open source bottle of building out these systems and solutions.

[00:15:08] And I think it's really important. I believe personally that opens openness is really important.

[00:15:15] I think the other thing that we need to think about, which isn't as talked about as much about, but it's sort of like comes from like, you know, relative to the sort of role in AI and DevOps and software development is, you know, there's also camps, if you will, about like, what is the focus that you're trying to build towards?

[00:15:35] Are you trying to build toward replacement right of a human or you try to build towards augmentation.

[00:15:40] And that's because AI is coming on this really interesting time where you've got a big contraction in the talent market around the layoffs around software engineers product managers, etc.

[00:15:51] And now you have AI coming in the mix. And is it going to be an efficiency gain or is it going to be a labor replacement and you know John Stewart had a really funny bit on his daily show.

[00:16:02] I don't know if you saw that, where he sampled all these AI leaders and they first said, Oh no, it's not going to replace us. It's here to, you know, it's here to help us and then everybody else. They subsequently said no, it's really a labor replacement.

[00:16:13] And I think that, you know, promoting an agenda of human empowerment, whether it's in the DevOps sphere or the software development sphere will cause us to build ultimately different things and we should probably be focused on that as as creators.

[00:16:28] I'm not worried for software engineers. Like you said, there's huge talent demand, but and there's no free lunch. We're going to have to continue to build systems to support this. But I think it's important to get smart on these technologies too.

[00:16:39] Yeah. And you took the words right out of my mouth. You must have seen the cogs turning in my brain now because I was just dying to bring up the John Stewart clip and the end game of AI. That's the big question, isn't it?

[00:16:52] Yeah. And you know, I think we still need more stronger security and cloud engineering talent and software engineers that are going to build all these systems and manage.

[00:17:00] I mean, we're going to be the ones who ultimately manage this era of evolution so we can adapt like we did in previous eras to it. I think better than other other industries.

[00:17:09] Yeah. And as an XIT support guy, I still get flashbacks of what happens with poor change management or testing, etc. But as workflows evolve with AI's incorporation, adding that into the mix, I think the role of testing and deployment is also undergoing a transformation.

[00:17:27] So how are you redefining these critical stages in the software development cycle? Because again, absolutely crucial. It's at the heart of everything, isn't it?

[00:17:35] It is. And you know, these were some of the same fears that people had when we were doing configuration management and infrastructure as code, right? So, you know, you always have to think about it in terms of push versus pull, right? Read versus write access.

[00:17:47] When you're using data to pull things out, for example, to tell me how my language framework works, pulls the internet as a chat assistant to help me write something, it's kind of low risk, right? If I've got it doing tab complete on code changes to my YAML, okay, but I'm going to have to do that.

[00:18:03] So if I'm going to have to do some code changes to my YAML, okay, maybe we're getting a little bit more, getting a little more risky. And then if you're going to putting it right in the middle of your CI pipeline, well, then, you know, that's getting a little more interesting, right? Now you have opportunity for injection of malware or other kinds of code changes that could potentially be downstream.

[00:18:23] So I think that's just, you have to kind of segregate things as things like I'm high confidence this push is going to work. I am it's a low risk thing, like I'm adding more test coverage. Or, you know, how do you think about how you incorporate AI to, I mean, to making changes in your either in your code base or your CI system?

[00:18:39] Like, would you put, would you let AI, you know, agentically make a hot fix on your production system? I wouldn't do that today. But you know, that's, that's something that you need to think about. And so the problem is, is that we don't actually do a great job right now of, of modeling like on a risk level basis how we make code changes, we don't really have exquisite systems for enumerating that. And I think objectively quantifying the risk of the action ourselves as taken by the AI is something that we need to think about.

[00:19:10] So I think that's one of the things that we probably need to think about as an industry.

[00:19:14] 100% and upskilling continuous learning. These are things that are becoming more crucial as the demand for AI and DevOps integration grows. And as I said a few moments ago, the importance of ensuring that nobody gets left behind in the AI world is really important.

[00:19:26] So how do you at AppMap support professionals in adapting to these changes? Any advice that you'd give to any individual looking at entering the, the, this area for the first time or an existing developer looking to upskilling the AI?

[00:19:36] What would you advise here?

[00:19:37] Honestly, the rapidity of change. So I think the big difference between cloud and the DevOps 1.0 cycle and the cloud is that we're not just talking about the data, but we're talking about the data that we're going to be using in the future.

[00:19:51] I, you know, I was off of X slash Twitter for a while, but I'm back on it because a lot of the conversation is going on there. And I think that's something that we need to think about as an industry.

[00:20:01] And I think that's one of the things that we need to think about as an industry.

[00:20:06] I think the biggest thing that we need to think about right now is the rapidity of the change of these models is significant. You literally have new models coming out, new methodologies coming out week over week.

[00:20:16] I, you know, I was off of X slash Twitter for a while, but I'm back on it because a lot of the conversation is going on there. Academic papers are being posted there.

[00:20:26] I think that's one of the things that we need to think about as an industry.

[00:20:30] So I think the big difference between cloud and DevOps is that you're not going to be using the same tools that you think are relevant for your job and just start experimenting with these, with the different platforms.

[00:20:41] Get to yourself to the level of comfort you can get to where you're using some of these tools yourself. You understand where the performance starts to degrade, where you, does it degrade with sequential uses over time?

[00:20:54] And so instead of actively using them, there's just no way you're going to be able to keep up with the, as an end user, there's no way you're going to be able to keep up with the change that you're going to need to do as an application builder.

[00:21:04] You need to know as an application builder for this life cycle of applications to come. You know, I think it's just, it isn't going to, you have to be both a consumer to understand the edges as a builder to be able to build better products.

[00:21:19] I think that's incredibly important, but gosh, I can't, I haven't read as many academic papers since I left my PhD program as I am reading now.

[00:21:28] You know, like it's really amazing that how there's a nexus of both commercial and academic work going on and in the, at the intersection point is where a lot of this innovation is happening.

[00:21:41] It's scary the speed of which technological change is going at the moment. And the reality is that it might not move this slower game as well, which just adds further challenges.

[00:21:50] I mean, it was only last year we were talking about chat GPT arriving and passing the bar exams, beating lawyers. Now we're talking about text to video and so many other exciting things.

[00:22:00] Looking forward, it has become almost impossible to predict the future because of that pace of change. But how do you envision the future of AI and DevOps integration, integration developing?

[00:22:11] And are there any other emerging trends or technologies that excite you or you just believe will significantly influence this trajectory over the immediate few months ahead and years too?

[00:22:23] Yeah, I mean, like you said, there's announcements coming every day. It's almost impossible to keep up with everything in the realm. I mean, Nvidia's announcement that you could run LLMs on a locally on a local machine with a preloaded board.

[00:22:36] I mean, that's one extreme right? Could things be getting so much smaller? And the other extreme is I'm doing $100 billion project to build the world's biggest super computer.

[00:22:45] I mean, everything is happening everywhere all at once. And for DevOps and developers, there's a lot of buy build versus build discussion going on right now in the AI and DevOps space because if there are a lot of companies that are thin wrappers on AI, big LLMs, and we've seen in the market it's really easy to replicate those apps.

[00:23:07] So I think we're going to have a Cambrian explosion of new apps both outside and inside current companies because AI makes prototyping applications so easy.

[00:23:17] But I think there are even bigger questions than that, that we as DevOps and developers and builders and makers need to think about which is some stuff, which is really what is going to be the input?

[00:23:32] What are going to be some of the new inputs that we're going to have? Are we going to have non-linguistic, non-internet type of inputs we need to sell for from a data perspective, from a throughput perspective?

[00:23:42] What are people going to want? The AI and chatbots themselves, they're kind of like the ultimate window into what people want. Kind of like a search bar was back in the cloud era.

[00:23:53] Are people going to have inputs for taste and smell and other chemical signatures so when you walk into a restaurant it's going to know how much salt you want in your food and adjust it?

[00:24:03] And how are we going to support those types of application experiences to deliver those kind of real-time adjustments?

[00:24:12] What is it that we're ultimately going to need to support? How are we going to think about supporting that from an infrastructure or application development space? I think it's crazy.

[00:24:21] And then I think the last question we need to ask is like, is this all going to be democratized or is this going to be the domain of a few companies?

[00:24:28] Is it a human right or is it a corporate service? And I suspect that people who will want to share more information than we probably know in order to improve their life and experience.

[00:24:39] But I think the question is with whom and what are they going to share and who will we be sharing it with?

[00:24:44] And as developers and builders, I think we have a bigger say than we recognize in this. We can choose to invest in open models, open source products versus closed models.

[00:24:54] And we can choose for transparency in prompts versus closed systems.

[00:24:57] And we can use our keyboards to vote with our consciousness and consciences and build the kinds of solutions we want to build.

[00:25:04] And I think ultimately that's we have a lot of power in saying how this is going to roll out.

[00:25:11] Yeah, 100% with you. And it reminded me of a story of my son recently. He's 23 now and he's doing the old travel around Europe thing.

[00:25:19] And he landed somewhere completely in the middle of nowhere.

[00:25:23] Kind of got the wrong internal flight, ended up in the middle of Turkey, Cyprus or somewhere like that.

[00:25:28] And as he pulled out his smartphone to look for what bus can I get and all things like that, he said all the front page of Google was all just irrelevant information.

[00:25:38] And he went to chat GPT of all things, chat GPT4 and he got all the answers that he needed.

[00:25:43] He got the right bus and it's almost changed the way he looks for information out.

[00:25:48] We both know that some of that information is slightly out of date because of the data sets.

[00:25:52] But he now finds it easy to get the answers that he quickly needs through that.

[00:25:56] And if that's where he is now, where are we going to be in five, ten years?

[00:26:00] The Internet as we know it's going to completely change, isn't it?

[00:26:04] It's going to completely change and we're going to have interfaces we have not yet experienced.

[00:26:08] Right. And so once we have those interfaces and we have those wants and desires and you have a merging of more kinds of models, right?

[00:26:16] Some of the models we don't even experience as end users right now. They live in academia.

[00:26:19] They live in biomedical research. There are all kinds of different inputs we're going to end up with as people who get to build and run these systems and design them.

[00:26:27] I think we have a lot of power in saying how we want them to be experienced and the kinds of principles we want to build into them.

[00:26:35] And we've had so much fun talking about the future where where AI is taken as DevOps and everything in between.

[00:26:43] But I'm now going to ask you, as we come full circle, to look back at your career, because I suspect you picked up more than a few stories along the way.

[00:26:51] So what's the funniest or most interesting story that has happened in your career that you are able to share with me today?

[00:26:57] OK, so I thought I was let me think.

[00:27:01] Some stories I think are only funny in retrospect.

[00:27:05] It's funny once you maybe get over the initial shock of the event.

[00:27:09] And this is a story and I'm not even sure my investors know this story.

[00:27:13] So I'm just going to share it with you like no one else.

[00:27:16] Just you.

[00:27:19] You know, I think it's a story about like what founders and entrepreneurs are willing to do to move the needle for their companies.

[00:27:26] You know, they say entrepreneurs are risk takers and those risks happen in like little and big ways every day.

[00:27:32] And this was a big risk.

[00:27:35] So we launched AtMap, my product, which is the personal developer centric observability platform at TechCrunch Disrupt in 2022.

[00:27:44] So we applied to the program on Alarq.

[00:27:46] We ended up not only in the top 20 of 2000 companies that applied, but we actually ended up in the top five pitching for the final prize.

[00:27:53] It was like a huge honor.

[00:27:55] It was like it's the Super Bowl of startups.

[00:27:57] It's a lot of pressure.

[00:27:58] You practice your pitch.

[00:28:00] It's five minutes, no notes, live demo and you get on stage for the top five pitch finale.

[00:28:06] And it's in front of thousands of people in San Francisco.

[00:28:09] It's a huge area in the Moscone Center that they do this.

[00:28:13] It's huge. Millions of dollars, millions of dollars of production value, millions of people online streaming.

[00:28:20] So anyone who's hesitant about public speaking like this is a huge jump in the deep end.

[00:28:26] And so I was feeling great.

[00:28:28] We're going to the finals.

[00:28:30] I'm getting ready to go on stage.

[00:28:31] I've got my hair and makeup done by the people backstage and I'm getting mic'd up and I'm five minutes to go on stage.

[00:28:36] So I'm going to get on stage.

[00:28:38] I practice my pitch.

[00:28:39] Thousands of people in the audience, millions of people online.

[00:28:42] And the production assistant turns to me and goes, uh oh, I broke your skirt.

[00:28:47] And apparently he had mic'd me through the back of my skirt and he had broken the zipper.

[00:28:53] And now it's staying on with like a tiny bit of thread.

[00:28:57] And I have no time to change.

[00:28:58] I don't have any clothes with me.

[00:28:59] So this is the classic nightmare scenario, right?

[00:29:02] When people dream about speaking in public, they dream about being naked on the stage in front of thousands of people.

[00:29:08] And so this is not a dream.

[00:29:10] So I'm having like one of these do or do not moments and I looked at the guy and I said, do you have any black gaffers tape?

[00:29:17] And my skirt was black.

[00:29:18] So he proceeded to tape me into my skirt and then paraded me up the ramp to stand next to my co-founder who was like, where have you been?

[00:29:28] Because we're going to get on in one minute.

[00:29:30] And I had like one deep breath and I walked down on stage under the lights in front of everyone.

[00:29:34] So this video of me is online.

[00:29:36] You can see it.

[00:29:37] Anybody who wants to watch it.

[00:29:39] The whole time I was pitching, you can think to yourself that I was standing there thinking if this tape doesn't hold, will I catch my skirt in time?

[00:29:47] Any second I'm going to be naked in front of thousands of people.

[00:29:51] What do I tell my kids who are watching this?

[00:29:56] Fortunately, my skirt did not drop.

[00:29:58] I did not have to explain to my investors, friends or strangers why I had a wallop drop mouth function type crunch.

[00:30:06] There was like a lot of camaraderie actually between all of the different contestants.

[00:30:10] And somebody actually said to me when I got off the stage, you're such a natural public speaker.

[00:30:16] And I was like horrified.

[00:30:17] I was like, are you kidding?

[00:30:18] I was terrified in ways you don't even, you cannot even imagine.

[00:30:22] But I was invited back the next year and I got to give a pitch on the same stage right before the finalists.

[00:30:29] And my delivery was probably not impacted because they thought I gave a good presentation.

[00:30:34] But like the second time I wore a dress with both buttons and zippers.

[00:30:39] So I couldn't possibly lose them.

[00:30:41] What a fantastic story.

[00:30:43] And with your permission, I will always write a blog post, a company in every single podcast episode.

[00:30:49] So with your permission, I will embed that video at the bottom of that blog post.

[00:30:53] Please do. It was a great pitch, I think, even though I was completely terrified.

[00:30:59] Oh, it almost sounds like a Silicon Valley episode there.

[00:31:03] I can't watch that show. It's too close to home.

[00:31:10] And finally, before I let you go, we've heard a fantastic story from your career there.

[00:31:16] What has been the soundtrack to your career or what song would you like to leave us with and add to our Spotify playing this?

[00:31:23] It's something I always ask my guests. But what would you like to add?

[00:31:25] Well, we actually do have an at-map playlist on Spotify.

[00:31:28] So you can see some of the soundtracks to our building work.

[00:31:31] But I happen to look at your podcast playlist and I am a huge fan of Blur.

[00:31:37] And I was surprised I did not see a Blur song on it.

[00:31:41] So I would love to put a song on your podcast list.

[00:31:44] And I'd like to pick the Universal by Blur because it's one of those songs you can crank up at your desk and sing to yourself.

[00:31:51] So maybe that would be the one I would add.

[00:31:56] Absolutely. We'll get that added straight to the Spotify playlist and quite a fitting song as well.

[00:32:02] If you're about to give a big keynote speech in front of investors and your skirt rips, it really, really, really could happen.

[00:32:10] Right.

[00:32:11] Really, really, really could happen.

[00:32:13] And for anyone listening that would just like to find out more information about yourself, your team, the work you're doing, AppMap.

[00:32:21] Where would you like to point everyone listening?

[00:32:23] Yes, I'm at Elizabeth Butler on X or Twitter and you can find AppMap and Navi AI on get AppMap or you can find us at appmap.io.

[00:32:36] Well, so much I love chatting with you about today from how DevOps principles are influencing AI implementation and vice versa.

[00:32:44] And also how professionals are adapting and upskilling to meet the changing demand, ethical considerations, redefining workflows.

[00:32:51] And not only that, we even had a hilarious story that I will remember for all the right reasons many years from now and even a great song to finish on.

[00:32:59] But thank you for sharing your story today, Elizabeth.

[00:33:01] Thanks so much for having me, Neil.

[00:33:03] For me, our conversation with Elizabeth today just illuminated that transformative role of AI in DevOps.

[00:33:09] Revealing a landscape rich with challenges and opportunities.

[00:33:14] And that journey from DevOps to AI DevOps is not just about integrating new tools or technologies.

[00:33:20] It's actually about rethinking the entire workflow, addressing ethical dilemmas and embracing a continuous learning mindset.

[00:33:29] And as we look ahead, it's clear the professionals thriving in this evolving ecosystem,

[00:33:34] they'll be the ones who can blend critical thinking with technical acumen, champion openness and prioritize human empowerment over automation.

[00:33:43] So again, Elizabeth, if you're listening back, thanks so much for sharing your invaluable insights.

[00:33:49] And for everyone listening, I invite you, my listeners, to reflect on how you can contribute to and shape this new frontier.

[00:33:57] What role do you see yourself playing in the AI DevOps evolution?

[00:34:01] What role do you see your business playing in this?

[00:34:04] And how will you navigate that balance between innovation and ethical considerations?

[00:34:09] This is a conversation we could talk about for hours, so let me know your thoughts.

[00:34:13] Tech blog writer at outlook.com, Twitter, LinkedIn, Instagram, just at Neil C Hughes.

[00:34:19] And let me know your thoughts.

[00:34:20] But that's it for today.

[00:34:21] I'll be back bright and early tomorrow with another guest where we will all together continue to push the boundaries of what's possible.

[00:34:29] But until next time, don't be a stranger.