Elizabeth Lawler and Pete Cheslock, AppMap – Improving Developer Experience

This is Textron TV. Hey everyone, welcome back to Tech strong TV. I really am I can't tell you how happy I am to have these two folks on with me today.

com and we've worked together and it's going on 10 years already and I didn't even put it together in my mind until about three or four months ago that they both are working at the same company. They've started up. I want to introduce you to listen.

Well, it's doctor Elizabeth Lawler if we want to be formal, but it's my friend Elizabeth and Pete cheslock. Who if you are a devops person, you know, Pete right? He's he's spoken it every devops days in the world.

I think or close to it and he's been a fixture in the devops. World for the 10 years. I'm doing this so Elizabeth Pete.

Welcome great to have you here. Thanks so much Alan for having us on Tech strong TV. Yeah, it's awesome to be here.

It's awesome to have you both. So look, I know you but the audience may not know you both Elizabeth. Why don't you give go first give us a little bit of your background and then Pete if you wouldn't mind with yours.

Sure. Well, you know as pick as you mentioned Pete and I are early adopters of new technology wave and we really like to get in on what's Happening next where the next Frontiers are in software development. And we're part of a new team here at @map.

Who's that's trying to solve some of the same problems through different lens that we were talking about 10 years ago when we were thinking about how we could accelerate software development and software delivery to help accelerate business. And I think that that's the common but I don't know maybe I'm speaking repeat. That's the common Bond.

I think that that goes through all three of us. Yeah, exactly. Yeah, I joined at map here in the last year.

I I'm a I don't know a constant startup junkie for lack of a better term. I keep going back to New startups. But you know this this pain around understanding the complex code bases when Elizabeth showed me what they were building here.

I immediately remembered to some of my earliest roles on operations teams where we have these undocumented untested systems and we're trying to keep them running like as an SRE or a devops person understanding what the code does so that we can maybe go fix something or put a patch on it just to keep it running and you know, what kind of flying these things blind in a lot of ways so being able to understand and visualize your code. It was very Primal, right? We these maps are very Primal and understanding you know, what the code is actually doing when it runs.

Yeah. Yeah, I mean, I think I think a lot of the things that you've been talking about Alan for like the last 10 years. There's a lot of what kind of feed and I remained still working on which is this notion that I Automation and visibility are like the secret recipes to Velocity and you know now you know, I think that the we've worked on a lot of the theory of constraints down the pipeline and now we're starting to Pete mention work on things that are constraining software development that is more at the beginning of the pipeline which is understanding the context of the software you're working on so you can make better coding decisions.

Yeah, I mean, you know one could say as much as things change they stay the same but we certainly are in a place of change, right? You know think back to 10 years ago the kinds of problems that Pete was working on and no one called it. That's sorry then right Google tied to put the book out yet.

But this the same sorts of problems that we're dealing with now except now, we maybe have a couple of different Tools in our belt. Right, probably one of the biggest, you know, I can't turn my head without seeing something about Ai and and chat GPT. And and it as much as I like messing around with chat GPT.

com Bubble Burst, right? I took a company public. I thought I was going to retire in 2000 and he didn't work that way right.

So when I start seeing like, you know, the guy at the gas station talking about Chad GPT, I start worrying that this sounds like a bubble or some kind of overhype. But it does this AI stuff does have the potential. To really change to change this game to change, you know, we you've been both been fighting this battle for a long time right?

Maybe maybe we have a new secret weapon do we or Just more, you know, Magic Bullets and Joe. Well, you know, I'm a big fan of generative AI in code development because I do think that it really does give you the same kind of boost and productivity that a code complete or does in your email. For example, I'm I'm happy to accept a lot of Gmail suggestions about how to set, you know, give us value mutation at the beginning of my email but it doesn't really know what the body of my thought is.

And so I think it's really good for anybody who is kind of like an early, you know who like text to, you know, a voice to text or who like code completion or who's interested in those types of the technology. I mean, I think it really does provide pretty accurate assistance and answers on demand and it cuts through a lot of complexity. It saves you a lot of Googling and it lets developers think about the problems at hand and use their brain rather than sort of slogging through.

You know artificial problems that they don't need to address but it has its limitation like you said and it's you know, it's a probableistic model. It's based many of them are based on language. So they don't really understand the sort of complexity of the greater system that you're trying to put the code into and that's really where you have to really think about what is function of this like just like writing an email.

It's more of a it's more than a series of syntactically correct sentences. Printed on a page. It's a thought it's a complete that and you have to look at these Technologies.

Like you're the editor. You're the curator. You're the person who's responsible for the design decision and the and the domain knowledge that all of those trees are going to need.

Yeah, I think what's really interesting is the tools are getting so much better that that developers can be productive. They're less novelty, right they kind of were novelty for a little while, but they're they're less. So now and developers can be productive with them for what I always try to call like things that are kind of like the the skeleton or the scaffolding of your code, right?

I need an API function to call Amazon S3 API I would normally spend time to like Google that or read documentation, but this can kind of autocomplete it and as Elizabeth was saying the thing where where we think at least this is gonna kind of be inevitable outcome of like easier code creation easier product creation is less understanding of the code that they just put in because technically no one wrote it right. It was just inserted into it and then also because that code isn't just gonna live on its own other things might call it now you talk, you know, this is the operator and me talking about distributed systems where now it's like what calls this function what does this function call a database? There's there's a lot more kind of happening in that stream and if we can kind of power those developers with some better understanding then AI gets real powerful.

Yeah, well two things number one Pete you head on something. I don't even realize if you if you realize what you there's a bigger issue, which is who owns that code when you took it at a church EPT or whatever AI you took them from do you own that code? You know because you don't want to be selling products where you don't own the code necessarily right?

And or someone's gonna come back and make a claim. Hey, that's not your coat. So that that's I think one whole area that, you know, I went to law school a hundred years ago, and I was taught that the legal system always drags.

The business world the legal system. I don't think is even an inkling of what to do about that issue right of who's gonna own that coat, but you know fundamentally though, you know, I remember interviewing Luke. Founder of puppet Luke Keen Kings, is it be?

Yeah canis this. Yes. 2015 and we would talk about you know, what is what's the ultimate game here?

Where can we go? And he and Luke said something to me. He said, you know by 2035 software will write software.

Right in 2015. It seems like 20 years in the distance who that I won't be here. Right?

But but you know what? We're not that far off right? We're what is it 12 years off and in 12 years look where we are today.

Yeah, that's not a bad guess. Not and not only is it writing it? It's writing test cases for it.

So, you know what? I mean? It's it's it is it is not only the writer but in some cases where people are not maintaining over side.

It's also the reviewer, right? So I think that you know, that's really where we have to start to, you know, understand the quality like you mentioned whether it's you know, whether it's a legal risk. It's a security risk.

It's performance risk, which Pete mentioned all of these risks exists in the database that generated the code. Right, and we're accidentally or intentionally pulling them in if we're not paying attention and I think that's really where the rubber is going to meet the road on the, you know, real business utility of using, you know generative AI on a really large scale inside of organization. Give give that you know, not everyone out.

Here's heard of that map. Let's really dig in a little bit give them an idea of what it is why it's important to our to our audience and how they can engage. Penis, you don't mind them.

Yeah, I didn't make people do it Elizabeth. You must see CEO thing, but good feet so app and what we're building is really all around helping developers understand the behavior of their runtime code and in how it actually executes. So seeing and recording the code it runtime the interactions.

Yeah API calls or test cases and things like that. And what we generate is inside the code editor insights whether it is some of our visualizations. We have automated sequence diagram creation, which a lot of Architects really love to be able to understand and use those design we can output and and generate open API documentation that's code accurate.

So you hear a big talk around code first API design, which is great. If you have done that from the beginning, what about all of the the code first people you have designed first, that's like Yes, we're gonna start but there's a lot of code out there that is not documented which which we help with and then this was actually something that lives was kind of touchdown. I'll let her dive into this which is really around understanding change because we if we are going to be bringing in a lot of code that no one wrote Some Humans are still reviewing this code.

So how do you improve that process around the actual reviewing of it? Yeah, I think if we're yeah to pick up where Pete was talking about. So now so we started with basically identifying a huge data Gap in the software creation process, which is that when you write software, there's nothing that shows you how it works when it runs, um, you know, any we don't do design documents anymore back in the old days you used to use.

Uml, right like you do anything you about thing and then you code to that the natural happened and we started coding first and then the system at the same time, we sort of diminished our emphasis on design work up front. We also increase the complexity of our systems. So many coders come into code bases that they're all and the only way the rate at which they can acquire knowledge or information about the code base is it's how fast can they read and how fast can they make a mental model that's accurate in a highly volatile environment that's like an impossible task to ask anybody to do and you see the topic of developer experience coming up around that which is like how much could we?

Really ask people to know in order to be able to make smart programming decisions. How much can we automate and make a parent so that they can just do a better job from the get-go because they have the right pieces of information. They may need to make smart choices and I think that you know, when we talk about trying to improve velocity, it's a balance of providing information keeping people in context and helping maintain flow and where does all that happen?

It happens in the code editor yet all this information about software behaviors everywhere. And so we said that's the Gap. That's what we want to go and learn solve that we're gonna do it through visuallyation and Analysis and provide those to those pieces of data in context with the actual code itself.

So developers can feel more effective and efficacious when they write code. You know, I will I'll add one last point because since all of us have spent a bunch of time in the SecOps world is the line and the SecOps was shift left right let the market just absolutely killed that term shift left we shifted left but we shifted not far enough. We shifted to the CI systems and we put this list kind of unnecessary burden on the developers.

We never armed them with tools that they could use so, you know, not that we're like security company because we do so much more but it's we have pushed and we've really tried to shift left really all the way to the developer in their code editor at code creation. So it's it's fundamentally the very beginning of the value chain, basically. Right and you know it's just so much if you can get that far left.

If we call it that it's just a lot easier. Well, it just makes for a less problems as you, you know, as that timeline goes. Let me ask you another question.

We're hearing a lot of noise about platform engineering. And how that kind of shows the ground for you know the developers in their applications. um Pete this sounds like something in your Valley work right in your Sweet Zone.

What what? You know, what where does that play in this? Hey, I I at the risk of starting to sound like the people I made fun of 10 years ago with the newfangled words for stuff and I'd be like haven't we been doing that forever?

I totally had the the kind of visceral reaction to like but that's what we've been doing this whole time. Yeah. I know we just it's a great term like it really does speak for you.

Yeah, what people kind of should have been doing which is building platforms. I often make this joke and it's not I didn't create this that the industry's been been trying to recreate Heroku poorly for a decade, which is that original first platform, right? It was so easy just commit your code and it was running there, you know, and so a lot of companies are trying to replicate that now you that's why you see the big rise in kubernetes, but the, you know fundamentally, I think from an operator standpoint or really opens up on on how the operators can spend more time again building these interfaces for their developers to to kind of Connect into their Location on the developer side.

I think it'll lead to inevitably a lot more product new product creation. It's kind of like the dream of microservices Might truly really land for a lot of Enterprises who will now have these these platforms that developers can just push to in a previous job. I was doing a lot of Amazon cost management we would talk with a lot of big engineering shops.

And the ones that were the most successful actually built full platforms for their thousands and thousands of developers so that the developers didn't even have to know what was behind it. They were just calling apis at the end of the day and it allowed them to hire these these just massive massive engineering organizations without having to scale up kind of operationally the same way. Get it.

Got it. A minute or two left. I want to make sure the audience gets the app map thing and why they should be going to the site and checking it out.

Let's talk about how the how appmap is sort of offered how people kind of engage. Elizabeth would you mind yes sure. I'm so at map is a you know free and open source code editor extension that developers can get right in their code editor.

So really are you know, our observation again was that observability and the notion of predictive observability was missing from this whole developer workflow. So we wanted to move all of that information all of that context that runtime Dynamic analysis right into the code editor with these beautiful interactive pictures that the developers love that is just right for what the code is that they're working on and then to provide in, you know, analytic tools and capabilities again, right in the code editor, which is where mistakes are introduced and Corrections can be made and so so yeah, that's how you actually get the platform. We do have some more Enterprise services that allow management teams to orchestrate and Analysis and creation of some of these Maps across larger systems and NCI and also report up about metrics of productivity and metrics of code efficiency and efficacy from the developer environment out and so so yeah, so it's really like going grab it and 40,000 developers have gone and grounded.

So we have over 40,000 groups who've been and you know who've installed at map from the vs code and generates marketplaces and our community is growing every day. We support Ruby Java Python and JavaScript applications. And so if you're coding on one of those and your team would like to have better alignment more visibility reduce the cost of communicating about code changes and improved code quality through better code reviews and Analysis.

Yeah, you can just go get it. That was great Elizabeth. He killed it.

Thank you so much. All right. Hey, we're gonna take a break.

io go check it out Elizabeth Peter two of my favorite people working together. It's a beautiful thing. I hope don't be strangers come back on.

Anytime you want keep us posted, okay? Thank you. Yeah, thanks so much always a pleasure you.

All right, we're gonna take a break now. We'll be right back.