Why Code Validation is the Next Frontier

Hey everyone. Welcome back here to Techstrong TV. My next guest is Arjun Iyer.

I hope I pronounced-- I think I pronounced it right. Arjun is the CEO of Signatot. And don't worry, we're going to tell you about Signatot if you're not familiar.

But first, let's have Arjun tell us a little bit about himself. Hi, Arjun. How are you?

Pretty good, Alan. It's a pleasure to be here, and thank you for this opportunity of introducing myself and sharing my story . You know what?

Thank you for sharing, right? Not everyone shares. So Arjun, I mentioned you're the CEO of Signatot, but as I said, we want to hear a little bit about your journey to becoming the CEO.

So why don't you share with our audience your journey? Absolutely. I've been the CEO and founder of Signatot for the past couple of years.

And before that, I was at AppDynamics- Sure ... which was a big name in the cloud native observability space. And I ran a few engineering teams there, which had one of the fastest growth paths within AppDynamics.

And then we eventually sold to Cisco, which was one of the biggest acquisitions at that time. Prior to AppDynamics, I've been in various startups, all related to cloud native and sort of more infrastructure oriented. So my passions lie in large scale systems, distributed systems, machine learning.

That's what I've been doing since I was in college. I went to college in University of Illinois at Urbana Champaign. Oh, sure.

And my thesis was in distributed systems. So that's kind of where I started. That was my passion right from the beginning.

And so my career has sort of followed that line of being inquisitive about how systems scale, how do they break down at scale, how do we design systems to be resilient and performant and highly scalable. So that's kind of been my journey, which led me eventually to AppDynamics. And based on what I learned at AppDynamics is the result of Signatot.

That's kind of why Signatot was born, because I've experienced challenges that developers face when building these scalable, highly resilient cloud native systems. And so that led to the birth of Signatot, and I'm happy to share more about that as well. Love it.

Arjun, it's funny. com. And AppDynamics, of course, was the very first sponsor- Oh, great ...

com back in the day. So Jyoti. Yeah.

And Steve, their marketing guy, Steve, I forget Steve's last name. He was the CMO there. But we didn't call it, of course, observability back then.

Yeah. Right? It was APM.

But yeah- That's true ... what a great company, what a great story that was. Look, I founded four or five companies in my lifetime.

Ventured backed three or four of them. And I've interviewed literally hundreds, if not thousands of founders. There's patterns that you see, things that you see.

Number one, the passion for what it is they're doing, it was what drove them to found the company. They had to be passionate about it. Number two, oftentimes they saw a problem that they themselves were suffering from, right?

They saw this in their last job. There was this issue that I just needed to solve, and then they realized that other people needed that problem solved as well. And so that's why they do it.

Other times, especially with serial entrepreneurs, they go out there and look for big problems- Mm-hmm ... and say, "You know what? " Jodie Ponzio from AppDynamics, and now of course Harness and all the other ones.

That's what Jodie does. I've asked, I've talked to him about this. Talk to me about what drove you for Signatot.

Where was your passion? What was driving you? Yeah, no, that's a key question.

And my background is in developing. So I am a developer by trade, and I used to manage a large team at AppDynamics, and this is something that really bothered me, where we transitioned to being completely cloud native back in AppDynamics. We started a little bit differently, but eventually shifted to cloud native.

And what I realized was the development workflow, the development life cycle for developers was not as smooth, right? They were wasting too much time on things that didn't matter or that could be done differently in a much more efficient way. So the whole software development life cycle, from the time you write code to all the way to you ship something to production, and how do you debug production, and how do you keep the system scalable, and safe, and resilient, had a lot of friction points.

And that is really the genesis of-- I'm very passionate about giving developers the freedom to experiment and to be very creative. That's really my passion. And I felt like developers were not given the opportunity to be creative because they were sucked into all these things that were just what I would call as chores.

Not really creative work. And that's really led me to Signatot, where the mission of the company is to actually give developers the ability to be free and to experimentExperiment and be creative, and especially developing production-grade cloud-native software. That's kind of the real mission.

And I'm happy to go into the specific problems, but at a high level, that was really what drove me to found Signidot. So when did you found Signidot? This was about four years ago.

That was when this was founded, and we raised a seed round of funding at that point. And now we have quite big logos of companies that are using it at scale. Some of them being names like DoorDash, Brex, Wealthsimple, Bitso, and other companies that use our software to really give their developers a very high-quality developer experience.

And so that's kind of the genesis and sort of the reason we formed the company. And yeah, that's our flag. That's been our journey so far.

And last year we had the biggest growth in the company's history. And so this year, that growth is accelerating even more because of the agentic sort of disruption that all of us are dealing with right now. Got it.

You mentioned it, so that horse is out of the barn, right? Four years ago, you didn't think you'd be here talking about what the role of agentic AI is going to have in your software development life cycle. But certainly, here we are.

Right? And I guess we should jump into it. It's hard to ignore that big elephant sitting in the corner of the room.

How is it affecting what you see in the SDLC, and how's Signidot changing its whole business as a result of this? Yeah. The birth or the advent of large language models and everything around it has been one of those completely disruptive technologies that very few of these I've witnessed in my career, but this is one, probably the biggest one.

And so because this kind of fundamentally changes software development, right now these LLMs are so good at writing code that it begs the question, okay, what is the impact? What are the roles of software developers going forward, and how can we work well with these tools? Right?

That's going to be the main thing. When creating code is not the biggest bottleneck anymore, because these tools do a fantastic job of creating code, but everything after that, how do you validate that code? How do you verify that it really works, especially in a complex cloud-native system, is where the bottleneck has moved to, right?

And that's exactly where we are positioned very well to sort of attack that problem. And it's exceptionally hard, especially for cloud-native systems. For smaller applications, it's less of a problem.

But when you have these distributed cloud-native applications with a lot of dependencies and a lot of components, just changing one part of the system, even though it's LLMs and all these tools doing it, it becomes extremely hard for these tools to completely validate and verify the system on their own, right? So that's something that they cannot do that well today. And so that's where the solutions like Signidot make it very easy for both developers and these agentic coding tools to be able to verify and validate the software at the speed of generation.

So that's kind of where we really are plugged in with these tools, is if you're using Cloud Code or Cursor or Codex or any of these coding tools, they need a way to verify what they've produced. And without that verification signal and the feedback loop to these tools, they're just code generation machines, right? They're not really shipping software to production.

So that's a very big gap that's there in the market today, and that's kind of where Signidot comes in, especially in the context of cloud-native solutions, cloud-native software. Absolutely. And talking about ephemeral environments, when I always hear that term, I immediately think containers, Docker, right?

Yeah. The average lifespan and that about ephemeral. I'm wondering, so in one respect, look, what better way to keep up with the ephemeral nature of cloud-native installs than with AI that could run at that speed?

Mm-hmm. Right? But does it present a human problem?

Because what happens is the AI can run so fast. The cloud-native infrastructure spinning up and down, the ephemeral aspect of it can run just as fast. So now these two things are running so fast.

And we want the human in the loop, I assume. We want human oversight still. How is a human to keep up with these two?

Yeah. No, that's the question of the century, I would say . Because it's going to be a huge topic of discussion.

Right. No. So software engineering is much more than writing code.

And that's sort of like what people miss when they jump to the conclusion that, oh, these write code, so we don't need that many software engineers. Like, that's a pretty big misconception and a misstep. You need even more software engineers because you need to understand how to design the system, how to architect it, what even to build, what not to build.

And that requires a multidimensional sort of thinking and strategy that only a developer brings to the table. These LLMs are not good at that. They are good at sort of following instructions and they are great token prediction machines.

And obviously when they generalize to a large number of the training data that they have, they can do an excellent job of producing really good code. But they need to be given the proper context, the proper instructions, and there has to be a human that's overseeing the overall development process. So that cannot be outsourced to these LLMs.

And you could argue that LLMs are getting smarter, so very soon they'll be smart enough to do even that. But that's sort of yet to be proven. That's sort of very far out there.

And so far, the evidence that I have seen is even simple tasks like write me an email, they don't do a very good job. I can write a good personalized email much better than LLMs, which tells me that they're far short of doing everything that a human can. So I think developers play a very vital role in that they own the overall strategy, and they own the overall design and decision making.

I think the decision making always will lie back and fall back on the developer because it's such a multidimensional problem. It's not just why the code is failing or why I need to ship this feature versus that. It requires so much of input and so much of context that only a human brain can sort of process and analyze that it's very hard to relegate those kind of things to these LLM tools.

So I think it's very early, but it's going to be a kind of a partnership that developers and other humans and other personas have with these tools that just amplifies the productivity of that human, but the human plays a very significant role in decision making and strategic thinking. Yeah. I've had these discussions with my own employees here at Techstrong because we're using agents a lot now.

Yeah. We've been using a generative AI. And I agree with a lot of what you said.

I think there's always going to be that human in the loop, and there's always, I call it the spark. The spark of creativity- Yeah ... is something that these machines are lacking.

Truth be told, I've trained, I've developed personas that they write in. It writes a beautiful email in my voice. It writes nice articles in my voice.

Yeah. I should say it crafts articles that I've written. It smooths them.

But I think there are some things that it's just not meant for, and one of those is that spark. Exactly. As I call it.

It's seeing the, wow, if I did this, I could do that. " Yeah. But I'm the one who saw and thought of it.

Where I see it most when I have it do graphic concepts. It'll come up with some concepts that are great, but it doesn't see the concepts that I see. It's the creativity.

Anyway, Arjun, did we mention the website? I don't think we did. Yeah.

com. That's kind of our homepage and our website. We have very detailed technical documentation also linked from that website.

So I would encourage anybody that's interested to go check it out. Absolutely. Yeah.

Arjun, I want to thank you for coming here on Techstrong TV today. It's been great hearing your story. I always like to meet AppDynamics people.

Great hearing what's going on with SignaDot. Come on and keep us posted. Absolutely, Alan.

Thanks so much for inviting me, and it was a pleasure. All righty. Arjun Iyer, CEO of SignaDot here on Techstrong TV.

We'll take a break. We'll be right back.