SOBRE ESTE EPISÓDIO
In a Tech Barometer podcast segment, part of a series of interviews with AI company CEOs, founders and leaders, Pendo CEO Todd Olson explains how AI is reshaping how teams work and how it helps tackle time-consuming, complex tasks.
Pendo’s Software Experience Management platform has helped companies like United Airlines, Morgan Stanley and iRobot optimize software performance by identifying issues, streamlining workflows and enhancing usability. To keep it running smoothly, Olson said his engineers now spend more time reviewing AI-generated code than writing it from scratch.
With AI changing not just development techniques, but also the dynamics of software teams, Olson is adjusting to a new paradigm and encouraging others to do the same.
“Everyone needs to rethink their role and how they can do it differently,” Olson said.
Find more enterprise cloud news, features stories and profiles at The Forecast.
Podcast interview transcript:
Todd Olson: How do you know… There are certain tasks we’re hardwired not to even try because they’re long and painful. That’s where AI can do things really, really easily. AI’s not eliminating all thinking that would be a scary and dangerous place for a lot of folks. It’s just shifting how we work.
Jason Lopez: These days we can see that shift with developers who spend less time writing code and more time reviewing what AI generates, while designers and product teams are using new tools to prototype and test ideas faster than ever. This is the Tech Barometer podcast produced by The Forecast, I’m Jason Lopez. Todd Olson is the co-founder and CEO of Pendo, a company based in Raleigh, North Carolina, which helps developers and product managers make their software more effective and intuitive to use. They do this by integrating into software solutions and AI has become a growing component of that. Olson spoke with us about how AI is reshaping how teams work, especially tackling time-consuming complex tasks.
[Related: HYCU CEO Simon Taylor Brings Order to Data Protection Chaos]
Todd Olson: I was talking about this with a few developers recently, is simulating user behavior. Testing can be very, very difficult, but creating real and simulated environments can be just prohibitively expensive and very, very difficult. But it could be very easy with AI. AI can just generate fake users, fake data, whatever. You give it some level of parameters and boom, it’s going to go off and go do it.
Jason Lopez: Olson says that hand coding systems for that kind of task would be much more painful. But with AI, a full simulation is far more efficient than manual testing, and opens up new ways to approach development. It expands what’s possible in the work itself and sparks new ways to collaborate for hybrid teams.
Todd Olson: We’re seeing team members share very openly, wow, look what I did with this. And then another team member saying, wow, I can apply that to my own work and get all this sort of productivity benefits out of it. I mean, a ton of our folks are recording Loom videos and posting it places and saying, here’s a video of me using AI to solve and experiment with this problem. Have a look. And people are commenting and experimenting. And I just think this is a really exciting time for a lot of folks.
[Related: Is Your AI Data and Infrastructure Audit Ready?]
Jason Lopez: That spirit of experimentation isn’t just cultural — it’s changing the work itself. Developers are finding that AI doesn’t replace their jobs, but rewrites the way they do them. Olson says it’s less about writing code from scratch, and more about refining what the machine produces.
Todd Olson: I was talking with our CTO and we were talking about thinking in terms of, yeah, if I had done this without AI, I could have written this code in say four hours versus AI is going to spit out the code. And then I’m going to edit, review and adjust it for maybe one hour. So that’s a, well, it’s a two hour savings if I’m looking at it that way, but it’s also a difference in skills, right? And you still need to have critical thinking skills because you need to understand if, if what the AI generated is correct, whether it actually meets the, the intent of what you’re trying to do.
Jason Lopez: AI isn’t taking human thinking out of the equation, it’s changing what we think about. Developers spend less time writing raw code and more time improving what AI creates. And, Olson says, people are rethinking their roles.
Todd Olson: We have designers, we have product managers, we have sometimes QA, sometimes not, but someone more focused in sort of the quality testing side. We have obviously developers themselves. I think we’re seeing AI sort of break down a lot of these walls and redefine what these roles actually mean. Like, like a really good example is actually in terms of designers, you know, designers historically, you know, obviously they, they would create everything from prototypes to they want to validate with customers. Now they’re using it via prompt. They can essentially generate an entire working application, you know, with, with, without having to, to lean on a developer or anything to that degree, much, much, much faster. And then get that prototyped application in front of customers much faster and get sort of much better feedback and much faster than they’ve ever been able to do before. And then when they’re collaborating with, with engineers, they can actually pass them a prototype and say, you know, Hey, this is kind of what I built. It kind of looks like this. Let’s tweak this. Let’s do this. Let’s make sure it’s scaled. Like it’s a different style of working.
[Related: Get Tech Insights to Move Faster and Smarter]
Jason Lopez: He also points out that to make real progress, companies need to give teams time and space to experiment, learn, and discover where AI can truly make a difference.
Todd Olson: We’re working on a project now to internationalize our products. That’s taking our products and making it available in different, different languages. That’s historically a very painful project. You know, you have to look for all the hard coded, essentially word string texts throughout your application, replace that with sort of an API call based on like, you know, geography. That’s actually a really good task for AI. And we’ve, you know, it, one thing from a change man perspective is, but if I went to my team and said, I need you to use AI for it. How long is it going to take? They’ve never done it before. They don’t know Adam. So what we needed to do is say, okay, we’ll give you a month to play around with AI and test it and prototype it for this, all this problem. Then tell us what it’s going to take now. So we gave them, we carved out time, we let them experiment. And then they came back and told us they can cut the estimate in half. Well, that’s a pretty good ROI, but if I hadn’t given them that time to experiment, they wouldn’t have actually known. So I think every company from a change, like they need to give their team space to try things out. It’s not just saying you must use AI. It’s give them space to like really play with the technology to learn from it.
[Related: Nutanix CEO Stokes Surge in IT Ecosystem Partnerships]
Jason Lopez: That kind of space can also depend on where the work gets done. And that’s increasingly in the cloud, where the newest AI models live and evolve.
Todd Olson: The cloud’s hugely important. Most of the foundational models are delivered via cloud infrastructure. So whether, I think the interesting thing to consider is that different models need to be better at different tasks. For example, most of my engineering team, when they’re doing any level of like code assistance or using anthropics models, cloud, especially a lot of the more modern, newer models. And what we’re seeing is the pace of innovation and change, like you want to be using the latest models. And like the models today, in some cases are significantly better than models six months ago. So like making sure you’re constantly trying to like use the same thing is sort of like top of mind. But yet for other roles, like non-engineering roles, non-code generating roles, we’re seeing OpenAI or ChattVT be the dominant decision. Again, both of these are delivered via cloud solutions. So I think cloud’s going to be important. Now, look, we have some customers that are very concerned around sharing their data with the cloud. So when I’m building products and I’m using these, I think one, you have to be very transparent about what models and infrastructure you’re using. Two, you certainly think about building in flexibility of potentially calling self-hosted models, which would be potentially open source models, think Metaslama or even things like DeepSeek. I think we’re seeing more people like think about leverage of those, which would be, again, you’re using a third party, but it’s open source and it’s hosted by yourself versus a third party hosted model. But I also think there’s gonna be opportunities to like, at least for us, use our customers’ models.
[Related: Turnkey IT Platforms Unlock Enterprise AI Success]
Jason Lopez: Many of Pendo’s customers have invested heavily in developing their own AI models, trained with their own data and context. The question Olson raises now is how they can design applications that tap into those customer built models instead of hosting everything themselves. But if there’s a guiding principle that emerged in our conversation with Olson, it goes back to his strong belief that we’re still in a wide open learning phase, especially human learning.
[Related: How Kubernetes Catalyzes Enterprise IT]
Todd Olson: We’re in the middle of our hackathon for our engineering team. Hackathon’s a week where they can basically work on whatever they want. But to stimulate AI activity, we altered how we do prizes. So we’re doing $1,000 cash prizes to teams who either build AI capabilities in some unique or interesting way or use AI on their projects in some unique or interesting way. And guess what? People are experimenting. People are doing things. We’re seeing interesting work being done. And yeah, look, money’s a nice little carrot, nice little motivator. But to me, it’s all about getting people to be uncomfortable and try something new. And I think you’ve got to find a way to be a catalyst for change at your company. And you’ve got to prioritize it. You’ve got to give people space to experiment. So give people space. Let them experiment. Even if the first few things they do may not quite yield the efficiency gains you were hoping for or may not be as high quality as you wanted, they are learning every step of the way. Encourage that learning and sharing. But yeah, I think that’s what you need to do.
[Related: Solving the Public Sector’s AI Agent Safety Puzzle]
Jason Lopez: Todd Olson is co-founder and CEO of Pendo, a company which provides software helping companies understand how their software is being used to help improve the product experience. This is The Tech Barometer podcast produced by The Forecast, I’m Jason Lopez. The Forecast published tech news and features both in text, audio and video. Check out more stories at theforecastbynutanix.com.
