Boosting Developer Innovation: Harry Wang on AI, Software Quality, and Security

Hey, everyone back here at Techstrong tv. I'm really happy to have back on with us today, Harry Wang. Harry is the Chief Growth Officer at Sonar.

I'm going to hear all about it first. Let's welcome Harry. Harry, it's great to have you back on.

How are you? Good. Good.

Thank you for having me again, Alan. Ah, it's a pleasure. A pleasure, Harry.

So Harry, as we were talking off camera, a lot of people here, the cheap Growth officer, and there, oh, this guy's a sales guy, couldn't be further from the truth. In your case, give people a little bit of a sense of your journey, how you came to be Chief Growth Officer, and it's kind of a unique position that you don't see at every organization. Tell us about it and why you are the perfect person for it.

Yes, and happy to share. Um, so Harry Wong here, I've been with the Sonar for a year and a half. I came from engineering background.

I was a software engineer by trade. And, uh, for me, chief Growth Officer is really trying to, at a company, a sonar, in this case, in my case, about leveraging, finding the levers, finding the growth drivers that could really help the company to grow, right? So that can range from m and a, from investment perspective, looking for adjacency spaces, looking for technologies and companies and team where we can acquire and make them part of sonar or can be actually internal innovation in the new areas, whether it's a security, whether it's actually ai, where we leverage our own internal r and d and then define a new product, define new market, where we can, we can grow and innovate.

Um, and then somewhere in between which we could really be actually working with the partners or working with our customers and really finding the new growth drivers. So it's, uh, I'm very excited about this spectrum of, uh, flexibility I have and my team have at Sonar. You know, if you think about it, it really is.

It encompasses what, what we used to call, uh, so I and I, back in the day when I was founding co-founding companies, I was always chief strategy officer, so I was responsible for corp dev m and a and, and partnership, strategic partnerships, that kind of thing. Yeah. Little bit of biz dev, right?

Yep. Yep. And then in, in your case, in like when I was Chief Strategy officer, also helped, had a hand in driving product strategy, go to market strategies.

Um, it, it really is a, a little bit of a, you know, jack of all trades, but you got your hands in really driving the company forward. And, uh, yeah, it's an interesting way of looking at it. Good for you.

Very, Very true. I mean, the, the title, which to me is really just actually a, a kind of a way of communicating what I do, what my team does. You know, a number of years ago in the, during the, the mobile transformation, we have people talk about, uh, growth hacking, right?

So, um, it's a little bit continuation to that where we're try and find growth. So we are looking for ways, different ways of finding growth and do what it takes to find that growth, uh, just happens in this paradigm. In the case of Sonar, we are really, this is sitting in between product and, uh, go to market.

Absolutely. Now, Harry, a lot of our audiences have, a lot of our audience have heard of Sonar. You know, it's a company that's been around, but not everyone has, and, and even a good percentage of those who have, may not really know the whole picture.

How would you describe what sonar, you know, the company and what you guys do, where you play, what problems you solve? Yeah, so Sonar is a Swiss based, uh, uh, company's been around for 16 years, uh, since founding we of the core of our missions to help, uh, developers, software developers to build better software and build it faster. And we provide a number of tools in the, as part of the CICD part of DevOps, as well as in the IDE as extension to refining quality issues as well as the security issues and providing recommendations on how developers can, uh, can, um, address those.

Um, and obviously in the last couple years with AI being introduced into the software development lifecycle, our mission is adapting. So we are not only helping software developers, we also helping the companions, the AI systems, and the AI agents out there trying to do a better job of helping the software development teams out there as well. Sure.

So, Harry, you, you, we, we made it three minutes and we got in ai, right? You can't help it AI's everywhere today. And, and there promises to be disruptive across so many different industries, but I would say that for whatever reason, an inordinate, an extremely inordinate amount of attention around AI has been focused on AI's ability to do coding to make our code better, to code faster, to replace developers, perhaps do better software testing.

Right? That seems to be such a huge focus of AI today. I mean, even absolutely philanthropic, the last version of Claude.

Mm-hmm. It was really optimized for co-generation. Yep, Yep.

Absolutely. Right? Yep.

Why the fixation on AI and, and software development, you think? Yeah, I think so. My, my own view of that is really it's a two part, uh, equation.

On one hand, AI engineers and scientists behind AI are the primary driver for this, uh, innovation. And, uh, you know, they, you, and we all have the tendency to look at the, for the areas where we are familiar with. In this case, uh, software development is very familiar, uh, to these in set of innovators, uh, behind the ai, right?

It is natural to start in that particular application area. And secondly, I think it speaks to the growing share of economy that's driven by digital transformation. And, and what is the underlying digital transformation in this case is really short software, right?

So if from a pure ROI, from a pure, uh, impact standpoint, it makes sense to start with, uh, software development as a key focus. And, and there's an amplifier you can do that. If you can do build software better with the ai, it means it transcend, it can translate into the GDP impact much faster than maybe some other functional areas.

Yeah, I mean, make no mistake. I eventually, AI is gonna disrupt and impact multiple verticals, disciplines, et cetera. But certainly, you know, software development is, is front and center right now.

Now, you know, we had this discussion, I was on text drug gang this morning, Harry, we talked about this Apple paper that came out. We were talking about it, but we also talked about living up to the hype, right? A lot of companies are saying, let's go small.

Let's play small ball with ai. Let's, let's, let's highlight some very specific limited type of functions or jobs that we're going to give AI to do. It, it, so it's not, you know, it's not an open-ended thing.

These are very specific. Yeah. And whether it's generative or agent Yeah.

Very specific tasks that lend themselves to ai. Yeah. Yeah.

Is that, at least for now, the prudent way to go about this? I, I think so, Alan, I actually remember our last conversation we're clunking about a distinction between gen AI and, uh, uh, gentech ai. It just actually, that was only seemed like two months ago, and things are moving so fast.

Uh, agent AI agents are not everywhere, right? Uh, maybe in two months, uh, from now on that will have a different conversation. Um, but I think the approach generally, uh, many companies out there taking is, uh, more measured approach.

com era where we actually seeing a dramatic productivity booth. We're not fully there yet, but I think the time for productivity will will come in terms of that improvement. Um, but taking, uh, embracing the ai, but taking a measured approach in both in terms of, uh, where we deploy AI in our work and, but also actually what type of a task going granular on what type of task we can delegate to the ai, right?

Um, the nature, they're actually, uh, widely, uh, wide conversations, uh, in the industry about how you choose the task as appropriations, right? There are tasks which is really, um, both actually high impact, but also quite easy to codify, think like in software development, think about unit test, think about documentation, think about issue remediation, let's say in the area of, uh, sonar, right? We did detect it, not only detect issue, but also helping our user to fix those issues, right?

These tasks, high impact, highly repetitive, and quite frankly, developers don't always enjoy doing that task, right? And also, it's a small tangible things, it's easy to reverse, right? So it's not like, uh, if an AI makes a mistake, it has the consequences on the actually the entire output.

You, you could, as an organization with the right measurement to whether it's right type of a assurance, the security to like sonacube, you could detect those mistakes and then reverse it easily, right? So I would say start with those tasks, and then as the organization build more confidence in ai, as the AI itself is actually evolving, you can start growing, expanding the horizon of deployment. It sounds like very prudent advice, right?

Because you give it a specific task, but at the same time, you also limit the blast radius, if you will. Exactly. If something weren't to work out.

Yeah. And, and to back it out. Mm-hmm.

You know, Harry, for as long as I've been in tech though, people always want to know, well, what's the ROI, right? And so in my background in security, it was near impossible to figure out the ROI because what you were measuring is something that didn't happen, right? If, if you, because insecurity, if nothing happened, you did your job.

Um, well, one could say that AI a little different, but how do we measure what, what's the right yard stick even to use to measure the ROI here? Yeah. I, I, I think that the, on the, as a whole economy, um, the, I remember, um, it was, uh, in, um, I forgot in which show, maybe da, that was in one of those conversations, uh, people talk about the prospect of ai, uh, for the society.

Eventually we need to see AI impact showing up, uh, in the GDP growth, right? So that's the ultimate litmus path, uh, for, for the economy. Um, are we seeing productivity gains as a society?

Are we seeing the GDP growth, right? Um, but specifically to a particular industry, a particular company, let's just take a software development as an example. Um, the ROI can be in dollar term.

It means actually, hey, are we building software with a lower cost or higher return in terms of, uh, the impact of that software delivered to the, the company's, uh, um, uh, revenue, uh, and then bottom line on top line growth, right? But oftentimes it's not necessarily a dollar term. It could be more macro measured impacts.

That could be productivity, could be happiness in, in the case of software, it could be developer satisfactions. It could be actually the turns on how fast you actually generating new features, delivering on those features and the quality measure. In the case of a sonar, we measure the quality of your code base.

We measure actually how many issues that you're able to address either by the development team or by the ai, right? So those, uh, leading indicators are also very important in addition to just simply a dollar ROI, uh, measure. Absolutely those.

And that again, good solid advice. Harry, if you don't mind, I wanna focus in on, on sonar. Yeah, absolutely.

We've, we've given advice to the whole world, but are you eating your own dog food or drinking your own champagne as some people like to say? How sonar using AI and, and how are you measuring its effectiveness? Yeah, it's almost like across the board.

Um, I would say from the experimentation standpoint, we are trying experimenting AI in every single function at sonar. Whether it's actually software development or it's our community engagement, customer support and marketing, you name it, right? Um, but I would say we do identify internally these several areas where we, um, put more wood behind it.

And, uh, that's a software development itself, not only to actually helping our sonar sourcers, uh, to, to be more productive, but also it's actually gave us more ground, um, more concrete how we build our own to better work with ai. So that shouldn't be a surprising answer. In software development, we use actually multiple, uh, AI tools as well.

Copilot, um, cursor. We constantly experimenting new market coming to the market, and that's some of that coming from our partners. Some of that was really just for commercial reasons we decide to use.

Um, outside software development, we do quite a bit, uh, experiments in marketing from content generation perspective, from a market research perspective, tremendously helpful. Again, back to the task I was talking about, these very concrete task, we can handle the, a collaborate with the ai, think of AI almost like our peer workers, right? But also in case AI is not actually delivering on the right type of, uh, output, we can easily reverse and that that doesn't hurt.

Um, and we can always having our marketer to really gate keeping the quality of the content and also for, uh, in strategy functions. In my own growth, uh, initiative, we often use research, uh, Google Gemini, deep research, fantastic capability in case, uh, you and your, the audience haven't tried. Highly recommend either deep researcher from Google or equivalent function from, uh, open ai.

They are all just actually, um, very, very thorough. I almost, you can treat 'em like a research assistant for a topic, uh, you can throw at it. They Really, I and It me too.

Yeah. And then also just, uh, in ana analytics functions, uh, finance, uh, sales operations, we start doing more, um, AI assist analytics. Uh, one example actually, I personally tried this really AI generat sql, right?

So to be able to look at the business analytics and, uh, generating business reports out of, hey, how, how we doing in online growth, uh, in the last seven days, uh, what is our returns? What is our new sign up? I no longer need to actually write a query myself or go to a analyst to do it.

I can just actually use natural language to generate those queries, uh, on the fly. It is, uh, it's a fantastic product they booster to, to myself and to my colleague Again. Me too, me too.

Imagine, especially on my side of the house now in media, this is, this is an amazing tool, amazing tool in that regard. But Harry, there's a dark side to AI too, right? Or potentially a dark side.

And that is the misuse of AI that certainly the bad guys in the security world are no strangers to that. But sometimes, you know what they say, the road to perdition is, is aligned with the best of intentions and even, you know, trying to use it for the right reasons. We can go off track.

How do you make sure that doesn't happen, Harry? It has the right level for policies and toolings, uh, uh, uh, at the starting point, right? So for, uh, back to the software development that, back to the DevOps, um, having the same level of policies and measurement safeguards we put in place, uh, for traditional software development lifecycle still applies.

But what we do need to look at at this moment is actually the sheer skill, the sheer volume of things that need to be vetted, right? So for example, in aerospace zone, we help detecting the quality and the security issues that in the past is really simple function of a number of developers in, in the size of the volume, the size of the code base. Um, but that's going through exponential growth.

When AI start churn through the code, the churn is going higher, the volume is going higher. At certain point, this is actually become very, very problematic just in terms of how you look at a different team have a different policy and some tool, some development team may adopt on queue, for example, other team in the organization may not, right? So it's very important for the company as an organization established what is the output, automate, safeguard and compliance requirement for software development and what's the right policy in terms of uh, what type of AI tools is allowed to use at what stage of the software development life cycle and what type of, uh, uh, verifications tools you have on the model on the software itself, and then finally through the test, right?

So having established that practice and being able to give a very clear guidance to the software development teams is a very important because ultimately that gave them the peace of mind so they can actually be more on the venturing side and taking more risk and then going faster. Um, and, and this, that life cycle, uh, to me is that is actually very productive in terms of it driving the increasing adoption of AI and not trying to restrict it. Excellent.

Alright. It's always the pleasure talking to you, man. 'cause I learned stuff, but I also see myself reflected in many of your views.

I think we share a common outlook at things. Maybe that's why I was a good chief, uh, strategy officer and you're a good chief growth officer. Um, but people wanna get more information about sonar and about what Sonars doing around ai.

Where would you send them? com. com, that's our, you know, homepage.

And, uh, we have, uh, you know, uh, constantly on LinkedIn, so follow us on LinkedIn as well. Uh, Katie and the colleague and the Irving partner marketing team is sharing our latest news pretty on a daily, pretty much on a daily basis. Great, Harry, I hope to see you soon and in person maybe at an event we're both at.

Until then, keep up the great work. It's sonar, come back and keep us informed. Okay.

Thank you Alan. I'm looking forward to it. Thank You.

How everyone, chief Growth Officer Sonar here on Textron tv. We're gonna take a break. We'll be back in a moment.