Perforce CEO Jim Cassens on Evolving DevOps Strategy Post-Acquisitions

Hey guys, thanks for the throw. We're here with Jim Cassens, the CEO for Perforce Software, and we're talking about how this company is evolving after multiple acquisitions, including some of the better known companies in the land of DevOps. But, um, there's a strategy and a method behind the madness, and I think Jim knows the answers to all these questions.

So, Jim, welcome to the show. How you doing? Good, Mike.

Thanks for having me. Really do appreciate it. And what a great topic.

There you go. So you guys have Puppet and I think BlazeMeter and a couple other folks and other companies that maybe are startups, but, um, how, in your mind, are all these things gonna come together and, and what might the future of DevOps look like in that context? Yeah, it's a great question.

Um, I mean, when we look at strategy around our portfolio in the companies that we've acquired, so we've done 13 acquisitions as an organization, we really look at smart integrations between products and how we can drive value within our customers, right? So you think about Delphix, which is, you know, virtualized data and being able to get data to customers that are testing, uh, efforts much earlier in the development cycle. And then combine it with BlazeMeter where BlazeMeter can generate synthetic data that can go with that production data that's masked to be able to fill in the holes during, uh, new application development, right?

Those smart acquisition drive value for the organization. So that's what we're really look at when we look at the portfolio and where we wanna put effort into, um, combining the different solutions that we acquire along the way. And then of course, we're just looking at AI and how can AI really transition, um, disciplines.

So when you take a look at our testing applications with, uh, perfecto and what we're doing there, the use of AI is really gonna change the way people look and how they perceive testing and what they can do earlier on that in the development cycle around testing, as we now start to automate the development of scripts around testing as well as maintaining those scripts and the ability that now to be able to really understand what's happening across a larger landscape in testing those applications than ever before. So we're really excited about being able to really, again, drive value within those organizations, allowing them at speed and scale to operate much quicker in the development cycle than they've ever done before. You mentioned ai and I wonder in the age of AI and especially agent ai, are the silos between the different platforms gonna become much, uh, more porous, shall we say, if not just completely disappear?

'cause I, I feel as I kinda watch this evolve that, um, I'm gonna have a task to do and I'm not really gonna care exactly what piece of software did it. Yeah, there is that, you know, people are worried on, on one hand saying, are we going to lose application developers, right? As, as agentic AI is now creating code versus having a human create the code.

And I look at it slightly differently than that. It's not that we're going to eliminate a developer position. What we're we're going to do is enhance that position, giving them the ability to work on some of the greater needs.

And really true fo truly focus on application development beyond just, Hey, I'm gonna create shells of programs, or I'm gonna write scripts that are gonna help me do certain things that can be done by the ent ai for us as a provider of solutions, um, to developers. We're in a nice position, we're gonna help them be able to manage the code that's being created by Ag agentic AI, potentially, right? Because this is gonna be volumes of newly created applications or newly created code that's gonna need to be managed.

It's gonna have to be validated, it's gonna have to be tested. We have the tools to kind of help with that, with that, um, speed that genic AI is gonna add to the velocity in your, in your pipelines. The thing that worries me as an employer and, and as we move forward with AI in general, when talent comes out of a university, generally speaking, they're doing the low level work, right?

They're creating some of those shell codes, they're checking those shell codes. If we take that over with Agen ai, how are those individuals that are just getting into the industry going to be able to learn at a higher level, learn how to be potentially an architect, right? How, like how are they gonna learn where all of the different pieces of an application reside?

'cause as you and I know, it's not as simple as saying, Hey, a database is just one database. They might be touching hundreds of databases to pull in the information they need. So I think as an employer, one of the things we're looking at is how are we going to train and develop this talent, the newly acquired talent, and making sure that they're prepared to be successful in an AI world, right?

'cause they're gonna need more advanced skills than they would normally get right out of the university. And to your point though, how much responsibility should the university assume to make sure that they're turning out people who can operate at a higher level? I think it doesn't come down necessarily to book learning.

It comes down to experience, right? So there's only so much a university can prepare someone for. That's generic across all industries, all development, you know, look at just the different ides and, and the different programming languages that you can use, right?

There's, there's thousands of those. So there's only so much a university can do. I think it's our, it's our responsibilities.

An employer once, once someone gets into the organization is to really upskill them and uplevel them to understand what are you gonna be doing here? Right? How is our environment maybe different than what you, what you learned at the university?

I think the university is gonna help them from, Hey, here's the new technology with ag agentic ai and other things that you can do out there. And preparing them to be able to hit the ground running. Like right now we're training our staffs in ai, right?

Um, so it's just gonna, it'll turn the focus from, I don't have to now train the, the newly acquired employee just outta university and ai, but are gonna have to train them across, Hey, how do you become an architect? How, how are things put together? Those will become meaningful and important moving forward.

Are there other areas that you're looking at that you think may make for a natural extension of your portfolio, either inorganic or organic wise, but, um, are there things like maybe, I don't know, more CICD stuff or that are part of the DevOps platform workflow that you feel you need to maybe own? Yes. So CICD is absolutely one of the areas that we're looking at.

Another one is anything around data. You know, when you start taking a look at data and how can we help people? Data is, it's the volume of data that's being created and trying people trying to manage this becomes harder and harder, especially as you're trying to deliver this earlier on in the, in the pipeline.

We've run into some regulation issues. You know, delphix will, will take production data, it'll size it correctly, so it'll reduce the size of what you need for your testing, and then mass that private data so that people can't see what's behind it, and then keep the referential integrity of that data. Well, for some organizations that are in highly regulated areas, they can't use production data.

They've been told and mandated by the organizations that we're not going to use any of our, our, uh, production data in any form from a testing perspective, which really creates the need for synthetic data. And that's what I talked about earlier with BlazeMeter being able to fill some of the gaps from a synthetic perspective. But we don't have an application today that generates volumes of synthetic data.

And I could see us either, either organically or inorganically acquiring the skills around generation for synthetic data, but then put it into the Delphix engine. So you can still mask it if you wanna mask it, you can still get that referential integrity if you're doing that testing, or you can subset the data into a smaller need for a specific use out of the Delphix application. So we look at that as a potential opportunity for us in the future.

I also look across, and everybody's talking about building AI apps, and when I look at those teams there, data scientists and their data engineers, and they're all engaged in something loosely called machine learning ops, ML ops. And then I look at DevOps and I go, aren't these two things ultimately gonna converge? And is that part of something you're thinking about?

We definitely are keeping an eye on it and, and we're inquisitive in terms of how would this fit into our portfolio? And more importantly, how is it play into the world of application development in the future? And is that what customers are looking for?

They're looking for a single vendor that can provide an integrated solution. Um, so where might it plug in nicely into our applications to provide that true benefit to, to the customers? Um, it's, it's an interesting play and there's a lot of different areas that we could go into.

Um, you know, one of the things you'll never see us do is kind of over overextend our bound. We'll look at those things that are, that are nice add-ins, as long as that's what the customers are demanding, that's what they're looking for. Um, but you won't see us jump too far afield from the DevOps, uh, solution base.

That's, that's really where we have our expertise. We wanna make sure we, we can maintain that defensible area of the room in, in our applications. You touched on this earlier, but I want to dive a little deeper on this particular point.

Um, we're gonna see a lot more code coming through those DevOps pipelines when I look at them. And I think one of the dirty little secrets of DevOps is the scripts and everything we use to create those pipelines are fairly brittle. And I wonder if, uh, we're gonna be looking at a point soon where they're just, the existing pipelines are just overwhelmed, and we're gonna need to think about that building DevOps pipelines differently.

Yeah, it's a good point. And and I think that's an evolution we're gonna see over the next few years. And, and I think, you know, there's only so much you can absorb as a, as a human in terms of the volumes of data that are being created.

And you know, as well as I do, it's gonna evolve over time, right? The learning engines of an AI are gonna get better at coding down the road. They might not be there today, although I do hear that they're generating some really qua high quality code.

But when you take a look at, you know, performance and you take a look at, you know, making sure that it's the code you're creating is optimal, that's gonna develop over time. It's not something that's gonna be there today. And that's what I mentioned earlier, that it's gonna have to be reviewed, it's gonna have to be, you know, how does this fit in?

How does the design look? How does it fit in from, from a more architectural basis across the entire application? Um, certainly it's just like anything else.

There's enough brittle code out there today, uh, in the world. It'll just be pushed harder and harder when it comes to, um, AI generating these things. And there's a lot of great fit.

I was talking to a customer the other day where they no longer felt like they had to write APIs. They could create the specs and give it to an AI agent and boom, they would get the API that they were looking for. Right?

Great use of ai. Let's, let's get the, the engineer really focused on the application at hand, um, versus maybe doing some of the, the, the remedial work that's needed for organizations and applications. I wonder your sense of what is it gonna be like to be a DevOps engineer in a few years?

Because there are, of course, everybody's kinda looking over their shoulder a little bit and going, well, who's moving my cheese? But there's another aspect to this thing. I think one of the other dirty secrets of software engineering is there's a lot of toil and a lot of stuff that we do over and over and over again that just, you know, it's, it's soul crushing.

So will we get to a point maybe where there's just more joy in software development because we're not gonna spend as much time and all that stuff work? I think that's true, not just a of software development. I think that's true of a lot of positions within the organization.

The person who's producing invoices over and over again, if, if they can just hand that work off to an agent and then focus their time and efforts on things that will change and drive the organization forward. I you're gonna see the same thing in application development. Some of those low level tasks go away.

To your point, the mundane, repetitive kind of grinded out aspects of application development are no longer needed within the organization and those individuals, 'cause now focus on the more meaningful work, the stuff like, you know, really being innovative in what they do and what they're delivering for their application. I also look at it from, you know, from a security perspective, vulnerabilities, being able to have AI do some things from a vulnerability perspective to make sure you're clean before you deliver an application to you and I through an, through an app on your phone. It just means that that app is gonna be a lot more hardened when it comes to the public than what it is today.

And I think that's a good thing. That's a good thing for, for us as consumers as well. Um, even around private data, you know, making sure that all of that stays secure and is, is neatly buttoned up in an application before it's delivered to the end user.

So I think, I think part of, you know, the world world I lived in originally where, you know, you had more bugs than you need, you had to deal with. Maybe a lot of that can be taken away with, with AgTech AI and some of the AI modules that are being built. And what is your best advice to the DevOps leaders out there today that are trying to navigate all this?

Um, you know, on the one hand I'll hear people talking about, you know, we're gonna build the next big software factory, but last time I checked, there's not many humans that are anxious to go to work in a software factory. So what, what's the right approach? Yeah, again, I think, uh, um, part of it is change is change, right?

And, and we're humans and we're people. And so there's a change curve we all go through as we are really redefining a role. And that's what we're going through right now is a redefinition of what application development really means, what these engineers are going through.

So for the leaders of those organizations, they're dealing with a lot of fud, fear, uncertainty, and doubt within their organization. Why? Because the developers are worried.

They're worried. Is agentic AI gonna take away my job? And how do I provide my f for my families?

How do I enjoy the life that I've enjoyed? What does it mean for me? How, how am I gonna navigate this change in the entire industry that's going on right now?

Um, so I think the, my advice to leaders is remember your people, right? Remember, they're humans. Remember, take a look at the change curve, understand where your individuals are in that change cycle and help them get to the other side.

And some of that can be through education. Some of that is making sure you're reassuring them. Look, this doesn't mean you're out of a job.

It means your job changes and you're gonna need to adapt to the changes that are coming. And then of course, I would encourage them to learn as much about AI as they possibly can. Right?

All of those aspects around ai, from security to what it can generate to the different models that are out there. Learn as much as you possibly can and that will drive security in what you're doing moving forward. Right?

Folks here in here, Hey, despite all the talk about AI and machines, it's still about the people. Hey Jim, thanks for being on the show. Appreciate it, Mike.

Thank you. All right, I'm back to you guys in the studio.