Join hosts Simon Maple and Guy Podjarny as they introduce AI Native Development, a transformative approach to coding. Discover how AI can fundamentally change the way we build software and learn about Guy's new venture, Tessl.

Episode Description

In this inaugural episode of the AI Native Dev podcast, hosts Simon Maple and Guy Podjarny dive into the concept of AI native software development. They discuss its implications, the shift from AI-assisted to AI native development, and how this paradigm could revolutionize software creation. Guy also shares insights from his extensive background in application security and software development, including his experience founding Snyk and his latest venture, Tessl. This episode is a must-listen for developers and tech enthusiasts eager to explore the future of coding in the AI era.

Resources Mentioned

• GitHub Copilot: GitHub Copilot
• Devin, the AI Engineer tool: Devin
• Tessl website: tessl.io
• Snyk Website: Snyk

Chapters

[00:00:00] Introduction and Welcome

[00:00:33] Defining AI Native Development

[00:03:30] AI Assisted vs. AI Native Software Development

[00:11:41] Future of Software Development with AI

[00:22:52] Introduction to Tessl

[00:26:13] Comparison to Snyk and Industry Impact

[00:35:21] Looking Ahead: Podcast Goals and Future Episodes

Full Script

[00:00:00] Simon Maple: Do you know, Guy, it's been over six months since our last podcast that we've recorded. Six months? Six months. Do you think we'll be rusty, or do you think this is going to be fresh and good?

[00:00:11] Guy Podjarny: We will have internalized everything, had time to think about it, and then not be able to articulate that at all. No, I don't. Right. Right. Should we start? Let's get going.

[00:00:33] Simon Maple: On today's episode, we're talking about AI native software development, what it means, why it's important and how it will change the way we create software today. Also, we introduce Guy Podjarny's new company, his ambitions, vision, and plans for the future. My name is Simon Maple. I'm the host of this episode as well as the show and joining me is Guy Podjarny, also a host for this show.

[00:00:54] Simon Maple: Welcome Guy.

[00:00:56] Guy Podjarny: Thank you, Simon. Hello, everyone.

[00:00:58] Simon Maple: So we've got a brand new podcast, the AI Native Dev. Let's talk a little bit about our backgrounds first guy, because of course, this isn't our first podcast that we've ever done, right? Tell us a little bit about your previous.

[00:01:09] Guy Podjarny: Yeah, this is,not the first podcast that we've run. the previous one is called The Security Developer. all name resemblances coincidental.

[00:01:16] Guy Podjarny: and it was all around how do we embed security into software development. I started it,eight years ago now or something like that. and over a hundred plus episodes had a chance to talk to some brilliant people in the, security industry, thought leaders in the dev space,really a great experience and it's still going strong with Danny Ellen as the host.

[00:01:34] Guy Podjarny: and at some point, Simon, you joined me in there as a co host.

[00:01:37] Simon Maple: Eight years seems crazy. I did a few years. nowhere near as much as eight years, but we had some good fun on that. We did a lot of the monthly roundups. And I interviewed a number of guests on that as well, but yeah, very interesting podcast.

[00:01:47] Simon Maple: also, shaping the movement of security into the developer motion. And onto pastures new here, we're talking about AI native development . Let's actually define a little bit about what that means because we don't want to, we don't want to start off with a buzzword that, that is meaningless and gets overloaded when we don't actually define it at its root.

[00:02:04] Simon Maple: What would you say is, what would you say, first of all, the need for AI native?

[00:02:10] Guy Podjarny: Yeah, I think,to get a little bit philosophical or historian maybe here, I think with every new development paradigm, really with every new technology that comes along, the first thing we do is we try to use that technology to improve the way we do things today.

[00:02:24] Guy Podjarny: and probably like a recent example of that is cloud. Cloud came along, you can, continue to run the VMs that you want, but you can do them on someone else's, environment, and that means you don't need to provision stuff. You don't need to,manage your capacity. You have a lot of those.

[00:02:38] Guy Podjarny: You could just make an API call and, I get the equivalent of your VM in someone else's data center. and so that was a great opportunity to lift and shift and get a bunch of boosts out of it. and, over time, of course, we've learned to appreciate that the true value, the true opportunity in the cloud is not in the lift and shift, which today we think of as, almost, negatively or a little bit as we say it in, with a detrimental tone.

[00:02:59] Guy Podjarny: But rather in cloud native software development, we understand that if you were to really change the way you build software, and embrace microservices and elasticity in your sort of infrastructure and,a variety of other practices, and DevOps really that evolved,then you can really squeeze the most out of it.

[00:03:15] Guy Podjarny: Yeah. and we call this cloud native software development. And it's harder to do it, right? we have the whole world of digital transformation, which is not entirely about cloud. but it is a painful process because you have to change how you do things. Yeah. And so I think with AI, if you look at what's happening today, I think we're doing, the same thing again.

[00:03:30] Guy Podjarny: And so there is, what I think of as AI assisted software development.and that is, using these amazing tools like Copilot and Test Generations. We're going to cover a bunch of those here on the podcast, to boost the way you develop software today. and the reason they're getting such adoption today, the reason they're so compelling is that they work within the system.

[00:03:50] Guy Podjarny: But at the same time, we have the belief, the conviction that,it is a good, path, but it's not the best path that the real opportunity, for developing software in the era of AI is to think about AI native software development. what is, the software development process that you would want to have if you were to think about it from first principles and really build the whole methodology.

[00:04:14] Guy Podjarny: Assuming, expecting AI and LLMs and all the power that they bring to be, to be available.

[00:04:19] Simon Maple: And so if we were to position it then with, tools and solutions that already exist today, obviously one of the big developer AI tools that, that kind of blew a lot of people away is GitHub Copilot.

[00:04:29] Simon Maple: Another one very recently, that was introduced, labeled the AI software engineer, Devin. how would you say these tools play in the space of AI native software engineering or software development?

[00:04:43] Guy Podjarny: Yeah. I think, first of all, they're amazing tools, right? And I, I highly recommend, in fact, I think it's critical for people to explore them and to use them where relevant, they're not without their flaws, whether sometimes it's like security issues with them, sometimes it is, hallucinations.

[00:04:56] Guy Podjarny: but I think they are absolutely a better way at least over the long haul to develop software and they supercharged it. There's a reason so many developers fall in love with them. I think of, let's maybe separate the two. so Copilot and coding assistant and a bunch of other tools.

[00:05:10] Guy Podjarny: I think a lot of them are really in the, the realm to me of AI assisted software development. And if you think about the, the requirements to be able to embrace those tools. the threshold is really quite low, right? You're already writing code, so there's no change in how you work.

[00:05:26] Guy Podjarny: and when you get code completion or autocompletion of any sort, then, you can eyeball it and you can see it's correct. So it's low friction, low, change for you. And so really all it does is it makes you faster. And so that's great. and so I, I am entirely in favor of, of those tools, and I think people should embrace them, but I think at the end of the day, that's, it's just selling short, the eventual opportunity of what AI can do.

[00:05:49] Guy Podjarny: These are assistants, they're not autonomous, which we can get to in a sec, but also again, you're still working there in, in the same methodology, in the same path, and I think when we think about. AI native software development, we're thinking more about, did you even need to write that code in the first place?

[00:06:04] Guy Podjarny: Could you,how did, the, insights that, Copilot had maybe to choose to introduce, that piece of code to you? Like that piece has never been communicated, right? What if you talk to that higher level language to say, This is the intent of what I'm talking about, and therefore have the system create those.

[00:06:24] Guy Podjarny: and once you really level that up, and we can dig into more, like what is the,the whole, like if you were to develop software in that fashion, what does that mean in, in how you build software, how you version software? and so I think Copilot is the right tool for you to be using today or like coding systems and not, Copilot, Tabnine, Augments, Cody,there's a million of them, not a million, but a bunch of them.

[00:06:45] Guy Podjarny: And, and I think you should be exploring the ones that you want to, that are right for you. But essentially, but it's just a, it's just assisted. It's yeah, they're giving you value today.

[00:06:54] Simon Maple: I'm still in my IDE. I'm still writing line by line. I'm reviewing what it does every single line and I'm doing things the same way, just hopefully faster or without having to refer to documentation and things like that because it's providing me what it thinks it needs.

[00:07:06] Guy Podjarny: Yeah. I think like sometimes when you when you level it up and you think of creating with AI, then it really works in these, one of these two modes, either the, human is. in control and the AI is the assistant, or, and maybe this is the case in the Devin case, the AI is the creator and the human is the editor.

[00:07:25] Guy Podjarny: And I think when you talk to, to most developers and most creators really, they didn't choose to be editors, some people love being editors, but maybe, many don't really. And so the assistant part is capping, is naturally putting a ceiling in terms of how much it would help, like you still need to be there.

[00:07:41] Guy Podjarny: maybe if you want to be a manager. On the other side, maybe if you want to be an editor. but I think there was an opportunity to think about how do you create with AI, how do you separate the pieces that you want to come from the human, and the pieces that you want to come, from the machine,

[00:07:55] Guy Podjarny: and so I think. I think that's oftentimes the opportunity with, with creating with AI. And, and there's a lot more to be said about that, but specifically for software and for software development. A lot of it comes down to the separation of this sort of intent from implementation.

[00:08:08] Simon Maple: Yeah.

[00:08:08] Simon Maple: And what's more fun, do you think? you mentioned a couple there. Yeah. From a developer, what gets a developer up in the morning? I've done, many years of development where I've actually really enjoyed You know, writing code, designing code. I get my aha moments. from a developer's point of view, a developer's role in the future, where's the excitement going to be for a developer, rather than just writing a short prompt to Devin or something like that, what's going to get them

[00:08:32] Simon Maple: creative and going in the future. I

[00:08:34] Guy Podjarny: think, I think it's a good question. And I think it's part of the exploration here as we look to, to figure out what AI native development is and what is an AI native developer, but I think really the drive or the two drives that, that make software development fun, one is the sense of creation.

[00:08:49] Guy Podjarny: You're coming in, you want a thing done, and you're getting it done, and I think that's very powerful. It's definitely, what motivates me as a developer. and I'm a rusty developer at this point. Rust developer or rusty? Rusty. Yeah.

[00:09:02] Simon Maple: Or dare I say, is there a difference?

[00:09:04] Guy Podjarny: I don't know.

[00:09:07] Guy Podjarny: I don't know how well I can read Rust. I was a polyglot developer at some point in my history before I became overhead. but, but I think. I think that the sense of creation is beautiful, creating, I don't know,when you create software, we were just talking, at Tessl, you're doing a bunch of stuff on the marketing front and automation, and, despite the show of affection you had for Python, as you were, building it,I think the excitement was from like, look, I've created this, right above and beyond whatever sort of practice you're doing.

[00:09:30] Guy Podjarny: and so I think that part is something that is critical for us to preserve when you're creating, you have to sense that you're creating, and that's the problem indeed on, of the editor. Yeah. It's if you're an editor, it feels like it doesn't feel like it's yours. Yeah. I think the second part that's interesting in, in software development is the,is the kind of the game-like mechanics on it.

[00:09:45] Guy Podjarny: When you're developing software, you can always level up your, you're solving a puzzle. There's a problem, you're writing code, and, you, you solve that problem. And then you can do that again at different levels, and you can always level up. and I think that's exciting, and it has even a bit of an addictive element to it.

[00:10:01] Guy Podjarny: It also, explains why, for instance, if you're an experienced developer, you don't get as much excitement, or you're not as happy to do things again, that you've already done, a hundred of these RBAC tests. setups, or a bunch of these login forms. And so that doesn't get you excited because it's like playing a video game when you're already at level 100 and you're being asked to do level 5 once more.

[00:10:20] Simon Maple: Yeah, and on that as well, there's probably a bunch of other things within the scope, not just coding, but the scope of software development, which actually when we think about creating a good whole, app or project. Things like whether it's documentation or testing and all these other areas that developers tend to put off, whether it's through, repetition or just not wanting to do them.

[00:10:42] Simon Maple: There are areas here that, that, AI can actually take some of those bits that a developer isn't as interested in or finds least fun, take it away and actually allow them to have more time doing these other things as well.

[00:10:54] Guy Podjarny: Yeah, absolutely. I think, creation is about the core, but the job of software development really is probably, for most people, maybe 5, 10, 20 % coding.

[00:11:06] Guy Podjarny: and all the rest of the time is everything around it. Now, some of that is still creation. it's whatever, like ironing out the sort of the quirks, it's testing, it's making sure that it works, but a lot of time it's indeed Toil, it's indeed repetition, it's updating the documentation to do, to say what you've actually modified versus naming it to doing twice, and a bunch of these things are things that can be automated.

[00:11:25] Simon Maple: So the way we're talking about it is, very much talking about the new role of software development. Presumably you're thinking, absolutely software development,from a, a career, a role, a job, it's absolutely needed going forward, despite the advancements in AI.

[00:11:41] Simon Maple: Is that right?

[00:11:42] Guy Podjarny: I absolutely believe software development is here to stay. I think coding is a different question. and I think it's again, part of that exploration. I think what is software development? Software development is bringing ideas to life in the digital world. It's around, creating something that is.

[00:11:58] Guy Podjarny: Either a repetitive task, or it is, aggregating information and activity that happens around, or it is a creative outlet, or just fun. and all of these things, like there's no reason to think that any of them go away. So we will still need to produce software. I think interesting things will happen to the software development world.

[00:12:17] Guy Podjarny: for instance, when you think about music, you think about, the democratization of creating music. It became easier and easier over time to. produce music, to creatce music And as a result of that, that changed the music industry. it brought a lot of new creators into the fold, a lot more people can create music, which is amazing.

[00:12:34] Guy Podjarny: It also created, a challenge, when it comes to finding the music that you want, it created this discovery challenge, it challenged a bunch of business models, and so things changed substantially, and so I think software development, the need and ability to create tools for the digital world, that would be alive and well.

[00:12:52] Guy Podjarny: I think core principles of software development, like the fact that a piece of software doesn't just get created and thrown away, but rather, you want to evolve it, you want to change it, you want to modify it over time,things like, Some pieces of software being more fun and disposable, while other pieces of software being, very mission critical and, attention to detail becomes very critical.

[00:13:13] Guy Podjarny: Like all of these properties, I think will be very much the same, but I think other attributes, like how do you build software? How do you maintain software? How do you collaborate around software? I think those will change. We have the advantage of only changing them for the better in terms of the ecosystem as a whole, but it does imply that some roles and some needs,will need to modify.

[00:13:35] Guy Podjarny: And I think what we're trying to do here is we're trying to,help define what that new world is, help preserve the goodness, help tap into the new power that, that we can provide and also help all of us really adapt to this sort of new reality, learn the relevant skills, discard some bad habits, etc.

[00:13:55] Simon Maple: And that's a lot of what we're going to do in this podcast as well, right? We're going to, we're going to, we're going to uncover the areas,of the AI environment, the AI,industry around software development. And we're going to talk to a ton of industry experts around

[00:14:08] Simon Maple: each of these specific categories within the AI tools and AI development.

[00:14:12] Simon Maple: which we'll talk about in a little bit. First of all, why don't you just,as a first episode of the podcast, tell us a little bit about your background Guy.

[00:14:18] Guy Podjarny: Sure, so my background, goes basically from being early in the AppSec industry and kind of building, tools for application security before it was known, through a bunch of acquisitions, went on to leave that and found a company in the DevOps space called Blaze, looking to make websites faster and getting acquired by Akamai, where I was CTO for a bunch of years,

[00:14:37] Guy Podjarny: and during that time period, really learned to, appreciate the sort of modern development and the changes, and how do you need to tune the tools of the craft, say developer tools, to the practices and how they're built out? post Akamai, which was a great sort of bunch of years, and I moved to London in the process. I, I left and I founded Snyk, which I'm probably most well known for at the moment, in which we indeed sought to help embed security into software development on the premise that,developers and development are moving faster and faster, that will continue. That's a good thing.

[00:15:10] Guy Podjarny: We love that. however, security, which is not. be able to keep up. And so you have to build security into software development. And we believe that you can achieve that by building a developer tooling company, not a security company. Snyk,is really my pride and joy, has gone very well, a journey with its own ups and downs, always, always amazing, but it's done well.

[00:15:28] Guy Podjarny: It's a thousand plus person company right now,a growing company that hopefully continues to be a generational company. and as the company scaled and grew, I got lured back into an entrepreneurial journey and, at the beginning of this year,made the leap to found another company, Tessl, which we'll talk about over here.

[00:15:46] Guy Podjarny: and I think there's a lot of learnings from those journeys on it. We're not going to belabor you too much, with those maybe except for comparisons to the AI native world. but I think probably the biggest,sense that I got out of it is that if you seek out, if you try to understand where the world is going and you try to, help push the areas that you think are correct, then you can actually help shape how things are done.

[00:16:06] Guy Podjarny: You don't do it alone. You do it with amazing thought leaders and early adopters around that make it work. but you can help everybody create like a platform for people to voice their opinions. You can help synthesize those. You can help voice and add your own. and I think a key part of that was building a platform for sharing those ideas, building a place for those that are at the forefront to be able to share their learnings, share their mistakes with others.

[00:16:33] Guy Podjarny: I think that was massive in DevOps. I think that was massive in developer security. and it's a core part of what we're trying to do here, on this podcast when it comes to AI native.

[00:16:40] Simon Maple: Yeah, amazing. My background's actually,as far, we have technical royalty with us here in terms of the industry, but my background's been, very much more in the pure software development for over a decade, really at IBM doing middleware development.

[00:16:53] Simon Maple: I moved more into DevRel space after that and I joined Zero Turnaround at about six years there. many of you may have heard, from the Java side of a tool called JRebel, which I was the, I built up the DevRel team there. And then of course we met in the Snyk days

[00:17:06] Guy Podjarny: You were already a Java champion at the time, so I'm not a champion.

[00:17:09] Guy Podjarny: I don't know what you're talking about really,

[00:17:10] Guy Podjarny: but

[00:17:10] Simon Maple: I don't have that in my name. I'm a prince compared to a founder. But,yeah. yeah, I joined Snyk of course. I had six great years at Snyk and, yeah, I, myself actually was doing a similar thing, wanted to join a new startup. And of course then I heard about Tessl and yeah, the rest is history.

[00:17:24] Simon Maple: But let's go back to AI native software development. as a developer in that, in, in that future that you just described, let's talk about the pillars of what AI native software development, actually means?

[00:17:35] Simon Maple: What does it stand for? What are the core things that when you look at something, you say, yeah, this is AI native software development in practice because it is doing this thing. Have you got a list or a few that you can share with us?

[00:17:47] Guy Podjarny: Yeah. And I think, first of all, I would start by saying we don't know.

[00:17:50] Guy Podjarny: Yeah. And it is a future. There's also a gap between the sort of expectation of destination, and where we are today, right. If you tried to build a service mesh with elastic computing and microservices in the very first days of the cloud, you would probably fail. So there's an element of the maturity of the technology and the maturity of the practice, and it goes hand in hand.

[00:18:09] Guy Podjarny: So some of this stuff is visionary, and some of it is things that can be applied today, and there's definitely a lot of,trying to predict and forecast and a key part of why we're having this conversation, why we're hosting this podcast is to, as a community, as a group, talk about these principles, talk about ideas.

[00:18:23] Guy Podjarny: And I'm sure whatever it is that I think I know today, will change, over the course of the coming kind of months and years. with that said, clearly we have some hypothesis. I think the biggest change that we should expect that I feel. I feel fairly convinced what will happen is the move from code centric to spec centric development.

[00:18:41] Guy Podjarny: When you think about how you develop software today, you write, you get some requirements, you write some code, you make a hundred decisions in that code that never leave the code, and the code very quickly becomes the source of truth, and we focus on evolving it and growing it, and the code couples what needs to be done and how to do it.

[00:19:01] Guy Podjarny: Andand because of that, again, like it, it all creates the sort of center of gravity and orientation. I think in the world of AI native development, we will have a means of specifying what it is that we need, and AI will provide the implementation. and that's a very,different reality.

[00:19:19] Guy Podjarny: First of all, it's not easy, like, how do you specify software? How do give the instructions that that you want, how do you express what it is that you intend to do? How do you scrutinize the decisions that the AI has done? How do you choose when to delegate, when to explicitly specify a thing versus when to delegate?

[00:19:37] Guy Podjarny: And so there's a lot to evolve around it, but I think fundamentally you will specify what it is that you want. and, AI will handle the implementation and because AI handles the implementation, all sorts of, interesting things will happen. First of all, of course, implementations will become easy and cheap to implement.

[00:19:52] Guy Podjarny: So this is akin to elastic compute, what happens, when you can generate an implementation on the fly. secondly, like once, once you implement on the fly like that, it becomes language agnostic. and that's interesting. Okay. I've built a piece of software here and it can be in JavaScript or in Python or in Java or in Rust, or in any of these languages

[00:20:10] Guy Podjarny: kind of right away. so what does that do to software ecosystems? How do we think about that? Where is it useful? Where is it not useful? And I think when you think about software development, that's also interesting. Okay. Like how do you, what is this thing? This sort of spec that you've created, is that a higher level programming language?

[00:20:25] Guy Podjarny: Is this basically now the source code that you're compiling, and if so, what is a compiler? Is there one compiler? Are there multiple compilers? How is there one spec language or the multiple spec languages? and then what's the right way to think about if software is so easy to create, what's the right way to think about, versioning that or packaging that.

[00:20:44] Guy Podjarny: If you consider the difference between, VMs or like physical machines that you host within a rack in the data center versus how do you treat VMs? Managing the, elastic machines that you just spin out and down, in,in a cloud environment, those are managed very differently.

[00:20:58] Guy Podjarny: Those you manage their life cycle, you manage their sort of, identity,their security posture, like all of these things, you manage them very differently. And so I think a lot of those types of questions, will come up from this sort of core change of moving from code centric to spec centric development.

[00:21:13] Simon Maple: Yeah. And I think that's going to be one of the biggest kind of, challenges for the new age of software development, where developers who are very familiar with, writing code and hopefully living as much time as they can in the IDE. to effectively understanding, okay, how can I actually address this problem through natural language?

[00:21:30] Simon Maple: Understanding the higher level concepts versus the lower level lines of code that they actually need to create. So it up levels,developers in their thought process, as to what's important for the application versus just, the specifics of the implementation.

[00:21:43] Guy Podjarny: And I think, Simon, like the first time we talked about this idea, then I think you said. that this will turn developers into architects.which I think I like that analogy. Like it is a higher level thought, maybe some will take the route of architects of this is how I want to describe the system.

[00:22:00] Guy Podjarny: Some might go more the route of the product manager. And maybe there's sort of other types of titles or responsibilities that we can't quite think of yet. But your ability to articulate what you want and your ability to articulate your, taste your preferences, the trade offs that you want to make, and of course the ability of whatever sort of system is there to help get those out of you and capture that, all those will control how good the result would be.

[00:22:26] Simon Maple: Yeah, and there's the potential there as well for us to, because we're looking, because we're thinking about it at that level, we're thinking about it as the business value, the, the use cases that we're trying to address rather than getting lost in the code. it gives the potential for building better applications, because we are not, we're not focused on the implementation, we're focused on the use cases and the higher level, functional and non-functional aspects of the application that we're trying to build. Apparently, this podcast is hosted by a company called Tessl.

[00:22:52] Simon Maple: Now, I've heard of some companies that like to omit vowels. Tessl is one of them. Have you heard of them before? I googled them before we started. There wasn't much there. I couldn't find it. No, I couldn't. I couldn't see that much. It's a condition I have. Oh, really?

[00:23:07] Guy Podjarny: I drop the vowels on the, uhthat and this sort of, irresistible drive for, for brevity, in the names.

[00:23:14] Guy Podjarny: yeah, so Tessl.

[00:23:15] Simon Maple: It's a bit better because Snyk had no vowels. So at least we've gone from, this is the zero to one, vowel journey.

[00:23:20] Guy Podjarny: It's an evolution, it's progress, right? Every, every company you get a little bit better. yeah, so Tessl is, the new company that I founded, earlier this year.

[00:23:29] Guy Podjarny: It indeed focuses, surprise, on AI native software development, our, goal, and I say our here because, Simon, you're helping build this dev movement here and this motion, and change alongside the company, our,goal really is on one hand, as we're doing here in this podcast, is help,promote this idea of AI native software development, help shape this with the community, help prepare us all into this next, generation of software development, and,build a platform that enables that, build a platform that is a sort of, built for AI native software development across its life cycle.

[00:24:04] Guy Podjarny: There's a lot of unknowns over there. We're still figuring it out as we're doing it. And that's the fun in it.

[00:24:10] Simon Maple: Yeah. Amazing. And today, of course, as this podcast, the day this episode,is aired. it's the launch of Tessl website, Tessl blog, Tessl the podcast, of course, as well. So an exciting time for Tessl.

[00:24:24] Simon Maple: what can people expect going forward?

[00:24:26] Guy Podjarny: Yeah, it is super exciting to be out there. And I think, this is, this is very much a mission that is anchored in the future, right, we're thinking about where is the world going? We're looking to evolve it with the system.I think what you can expect from us is just a lot of collaboration, a lot of, conversation and a decent amount of experimentation on it.

[00:24:46] Guy Podjarny: we hope in hopefully even this year,to launch some products that you can try in beta form. You can give us the opinion on. you can work with us to, again, reshape and figure out what is correct. get a little bit of mileage of what actually works with today's technology, what doesn't.

[00:25:04] Guy Podjarny: Get that feedback to whomever it is that needs it and adapt ourselves accordingly, and so I think in the meantime, what you will see from us is really a lot of this. We're going to, publish content, we'll talk to smart people, and help give stage to their thoughts. we'll talk to, you know,

[00:25:19] Guy Podjarny: AI tools that you can actually use today. So a lot of the things that we expect to do in this podcast and you'll see on our blog, you'll see in our newsletter,it's just really boosting the conversation around this domain. And in the background, we're working on trying to build a product platform that learns from those and that helps both fulfill some of these promises, but also,try out some of these ideas and see what works.

[00:25:43] Simon Maple: Absolutely. And I think one thing I will say, of course, is,this is a journey of collaboration. you're more than welcome to go to the tessl.io website, where we do have the blog and podcast, do sign up there as well, and you'll hear not just about the great content that we're pushing out, ourselves and with others in the industry,

[00:25:59] Simon Maple: but yeah, as, and when those,new products, become available and open for early stage testing, we'll absolutely let everyone know, and, hopefully get you on board as early as possible for your feedback and to,see what the next gen of software development hopefully looks like.

[00:26:13] Simon Maple: now we drew the comparison actually with Snyk a little bit earlier and how Snyk was really there to, address the change in the movement of bringing security into the, into that developer fold. How would you say, how would you draw that parallel with Tessl, doing something similar?

[00:26:29] Guy Podjarny: I think, Snykgave me the conviction that, if something is the right motion, then if you really explore how to make it happen, it can become reality, and so when I started Snyk, a lot of the pushback was, Developers don't care about security, they will never embrace security.

[00:26:50] Guy Podjarny: And there was really no, nobody arguing with me that they must, and so there was this oxymoron of, on one hand saying, yes, we agree, DevOps, cloud, all of these technologies have accelerated software development. There's no way security can keep up. They're already falling behind. we don't see another way, it has to get in developers.

[00:27:09] Guy Podjarny: But developers will never embrace security. They're, they're all sorts of like cynical comments. They only care about their sort of feature function and things like that. And I think the, the sort of a cynicism, isn't very helpful. What we've done, it's neat, cause I'll actually, If it must happen, how do we do it?

[00:27:27] Guy Podjarny: And we really embarked on that. This is okay, if we only cared about getting developers to embrace security, what would we do? and that's the path we pursued, and I feel like we've been successful. We're far from perfect. Developers have not like generally see security as their sort of, top of the list, but I think we've, managed to pioneer approaches to how to get developers to, to make security a natural part of their daily flow.

[00:27:47] Guy Podjarny: You now see most dev tools, introduce some element of security into them, and you've seen, pretty much all security teams, embrace this as part of their security program. And I think if you stick to the trend line, if you stick to the destination of things that must happen, then you can zig and zag, it's not like we got it all right, but you can actually there, there's real opportunity to help find that path.

[00:28:10] Guy Podjarny: And when you do, once you open that door,you get to shape the industry. Like you get to, to modify or help unlock a certain practice. and so I think that was,very, uplifting,and above and beyond whatever the success of the business of it, it really felt like, we managed to,to put

[00:28:28] Guy Podjarny: this slice of the industry on the right path. I think our ambition here is bigger.It's not the developer security, is small by any, stretch of imagination. but I think here we're talking about something, a bit more. a bit more massive. and I think part of our challenge is to think about how do we narrow this down?

[00:28:46] Guy Podjarny: How do we bring it down to reality? How do we make it work today so we start learning because like everything, everything is a product and v1 always sucks. And so so you have to get started and I think the kind of the, maybe the tactics that are quite critical to, to making that happen is the first and foremost, find your allies, figure out who is also on this sort of bandwagon of AI native software development, get their thoughts, try things out.

[00:29:11] Guy Podjarny: We're keen, like what we're building is very broad. And so because of that, we can't ship it quite as quickly as we ship Snyk, but we're keen to put it in the wild. to get the feedback, to be told when we're being idiots. but I think you have to work through iteration.

[00:29:22] Guy Podjarny: You're not going to get it right. and then to stay true to the conviction. And so my conviction in Tessl, I think our conviction as a company is that AI native software development will happen. That AI is too dramatic. The ability for a machine to generate code. is so powerful that there's no way that the correct way to develop software is the same, as it was today.

[00:29:43] Guy Podjarny: And basically, if that premise remains correct, then I think there's value to be had here and there's valuable kind of product and solutions to be created.

[00:29:52] Simon Maple: Yeah. When you say it's, I love the words, it's a bit massive. It's a little bit massive that there is, everyone's mileage varies today with AI,

[00:30:04] Simon Maple: when we think about this, such a large jump, is this just a fantasy in terms of where the future versus the actual technology that we have today. Is this, is it, like you said, a couple of times, hallucinations, LLMs making things up. Are we where we need to be right now?

[00:30:20] Simon Maple: How do we get to that stage where we can trust AI enough to actually be comfortable in this land of AI native software development?

[00:30:27] Guy Podjarny: Yeah,I think it's a very good question. And I think,practically every core technology that we use today and, technology advancement, was perceived as a fantasy at one point in time.

[00:30:38] Guy Podjarny: I think it's less of a fantasy than, some things, whatever, flying cars, it's not a sci fi movie. I do think it stretches the boundaries and in general, like, LLMs are, are limited today. if you talk to anybody that is building LLMs, they talk about how they're so weak still today, because they see all of that potential.

[00:30:58] Guy Podjarny: But, I guess we're building Tessl and we're building this, or, trying to develop this notion of AI native development,in a way that is anchored in the future, and so the bet we're making is that these models do become orders of magnitude better, that our ability to wrangle them, really improves as well.

[00:31:15] Guy Podjarny: So fine, there's the ability to generate code, but not everything has to be AI, not everything AI has to be LLMs. And so these are tools that we can create. and yes, it's on us to try and find,the, narrower paths of maybe slightly simpler software or, more achievable,solutions that we can build, in the short term, but fundamentally, I guess we're building the company with an eye to the future.

[00:31:38] Guy Podjarny: And so the criticality isn't about how quickly. AI native development comes about and more about, what's the best way to get it right. what is it that we can do to learn? And I guess we we have the, the advantage, with the sort of the background of the team, of having a little bit more, string, a little bit more rope,to take a bit of time, to develop that.

[00:32:00] Guy Podjarny: So we want to develop it with everybody here. We want to,ship things, so we're told that they, that they're wrong, so we, know which paths to take, we want to, try things and see if we can make them work. My expectation, I don't know if it's prediction, but at least, what I think is good for plan A is that the rate of progression for these LLMs and specifically code generation within them,would be very fast. it, something that people don't think about too much is code, like a small number of other domains, is verifiable, and so when you think about how much can you battle hallucinations and things like that, the more kind of, elusive the domain is, and it's hard to tell whether a result is correct or not en masse,

[00:32:46] Guy Podjarny: the harder it is to get rid of hallucinations. but code is actually a place in which oftentimes you can get deterministic results, and as a result of that, it's actually, probably maybe the, or definitely one of the fastestprogressing domains within the LLMs. There are also some dedicated,foundational models that focus on code only. and so I think there's a lot to be, that is coming quickly.

[00:33:08] Simon Maple: Yeah, amazing. and I think, when we think about the structure of this podcast, we're going to be going forward doing like two, I guess we're trying to, we're trying to do two paths here. One, which is the today looking forward.

[00:33:20] Simon Maple: We have this vision of what the future Could look like, but we equally have these tools here today and how can we improve our existing software, development practices today? Every week on a Tuesday, I'm going to be, releasing a new podcast episode, which is going to be looking at the tools of today.

[00:33:36] Simon Maple: whether it's,development, through code completion, whether it's testing or documentation or whatever it's going to be, and we're going to, we're going to be talking with a number of different people,to really understand what we can do today with their tooling, why and how AI improves that aspect of software development and what they see are their next steps in this, AI journey going forward.

[00:33:56] Simon Maple: You're doing a little bit different, of course. you're starting out, like you say, anchoring in that future, who are the types of people you're going to be talking to?

[00:34:02] Guy Podjarny: Sometimes it's the same people that you might be talking to or people from the same companies, that are just trying to think further ahead or out there. and I think many of the conversations that.

[00:34:11] Guy Podjarny: that you'll be having talk about today. But a lot of the technology and practices that are being built in this AI assisted path are actually critical learnings to be able to figure out the destination. I think some of the best companies,in the sort of the AI dev tools space, they don't just think about the thing they'll develop today and how do they, whatever, beat the competition,

[00:34:31] Guy Podjarny: but they also think about, what is the destination? How does that, work in five years time. What did they imagine? And oftentimes they will have unique insights about that domain. I'll also try to get a bunch of people that come from different worlds of creation with AI, of building AI tools, and try to kind of capitalize and leverage their learnings, about where the world of AI native other things, is going. Because I think in many of these cases, those things apply back to the world of development, and we want to be inspired from outside the industry as well, and I think the combined views

[00:35:06] Guy Podjarny: of, of the today and the tomorrow, should help us try to draw a line to how do we get from, reality today, which already sometimes feels like science fiction in its own right, right? When you go off and, and chat to one of these LLMs, to, to the sort of, and I guess further, future.

[00:35:21] Simon Maple: Yeah. It does feel like. science fiction, actually, because, me writing Python, who'd ever have guessed that, right?

[00:35:26] Guy Podjarny: But is that dystopian or utopian?

[00:35:28] Simon Maple: Who knows? Could be either. so yeah, one of the interesting things as well is that, we're not here to, we're not here to talk solutions. We're really here to ask the questions and learn as a community.

[00:35:39] Simon Maple: so we want to like really understand what the right questions are to be asking the right people and try and understand this new world as a group and as a community.

[00:35:47] Guy Podjarny: I think,the value of getting these opinions from the different peoples really cannot be understated.

[00:35:52] Guy Podjarny: we're deep in this domain, we think we'll have some things to share, we'll have our own learnings. but really, and again, this is maybe a learning from the developer security journey. the true value comes from talking to many others that have used, AI dev tools, that have built AI dev tools, that thought about where they're going, to, that, that are actually trying a bunch of these paths.

[00:36:09] Guy Podjarny: And in those conversations, sometimes you get nuggets of wisdom of hey, this is an entirely new idea that you didn't get, and those are very valuable. Sometimes you get affirmations for things you already know, and sometimes you get pattern matching, you get, okay, everybody is doing this, okay, there's something there.

[00:36:24] Guy Podjarny: That we have to understand, so like with anything data,there's elements of quality, you're going to get these sort of moments of brilliant from, brilliance from, these amazing people around, and there's going to be...

[00:36:34] Simon Maple: My episodes!

[00:36:38] Guy Podjarny: Well, I'm not talking about brilliance of the host.

[00:36:39] Guy Podjarny: I'm talking about the guests you're going to have are going to be brilliant. and there's going to be moments in which you you're seeing, the,the quantity, and I think the benefit of many years of the secure developer, and we've talked about that many times, also allows for the, perspective of trend, which I think will be fascinating.

[00:36:55] Guy Podjarny: I, I bet it would be super, super interesting to listen to some of our early episodes, maybe with a little bit of cringe about ourselves, on it, but see what have been the sort of the perspectives at the time and listen to them two or three years later, and, my, my expectation would be is the, our understanding will be like an order of magnitude, more evolved.

[00:37:14] Simon Maple: Maybe two or three months later, we'll see.

[00:37:16] Guy Podjarny: That's fair. I don't know if we'll learn that fast..

[00:37:20] Simon Maple: But talking about what's next, yeah, we've got a few episodes coming up already. In fact, the person that I am speaking with next, Amir Shevat, I know someone you've gone back many years with as well. One of the things that we're going to be talking about is categorizing AI tooling.

[00:37:34] Simon Maple: So what we want to be able to do is say, okay, what do we deep dive in? We don't want to just say, let's deep dive in AI, what do we deep dive in, and so we're going to be categorizing the various AI tools, developer tools to understand what are the aspects that developers or organizations needs to be thinking about and understanding where they can apply, AI into their organizations, into their development organizations.

[00:37:55] Simon Maple: And then each of those sections, we're going to in those. We've got Amir, up next, who I'll be talking with, and then we'll deep dive into our first topic, which is going to be around, AI code completion.

[00:38:05] Guy Podjarny: Yeah, and Amir will be brilliant, on, categorizing those, and Amir has done so many things in dev, in dev relations, in dev movements, including,building out a lot of that practice at Twitch and at Slack.

[00:38:15] Guy Podjarny: And now is an investor in really smart, AI dev tools. So I'm very keen to hear that episode.

[00:38:20] Simon Maple: Can't wait. Can't wait. Excellent. until next time, thanks everyone for tuning in. And Guy, it's a pleasure as always. Yeah.

[00:38:25] Guy Podjarny: It's great to do this opening show with you, and I'm really looking forward to a whole horde of amazing conversations.

Podcast theme music by Transistor.fm. Learn how to start a podcast here.