.jpg)
|
July 30, 2026
Technological development is having an impact on every aspect of our lives, and AI is among the most significant innovations that has forever changed the way the world works. While some treat it as a threat, for others it is an opportunity and a step toward a better life.
These two perspectives lead to one fundamental question: how can we adjust to the changes that AI brings so that we can use them in our advantage?
To discuss this topic, we spoke with Peter Leibiger, the new Chief Technology Officer at FIO Systems AG. Peter is an AI specialist who helps connect technological innovation with business needs, creating practical solutions for the financial and real estate industries. His experience allows us to explore deeper topics such as embracing AI and the changes it will bring, finding the balance between innovation and tradition, and navigating the uncertainty that comes with it.
Thank you — though I've known FIO for seven years already, from the outside, building AI solutions and mobile apps for them. So this wasn't a leap into the unknown, more a change of perspective.
What makes it interesting is the question behind the position. AI is everywhere right now, but how do you get it into a large B2B application in a way that actually helps users? And how do you open that application up, modernize it into an ecosystem others can connect to? We're not trying to build AI features so we can put a label on them. We want to solve real problems, and sometimes AI is what gets you there.
And there's a team here that knows the business inside out. I'm looking forward to what we build with that.
The number in the title is real, but the talk wasn't about the number. It came out of a problem we had. Part of our system kept getting slower over six weeks and we couldn't find the cause. We were missing the telemetry, the data, the context. Once we had that, the fix was easy. AI helped a bit too, but only once it had something real to work with.
But that's not just an AI story. Missing context is why people go in circles as well. Give it nothing and you get guesses back, give it the real picture and it gets useful very fast. Same as with a colleague.
That's also why I wanted to share it internally. Making the dark corners of our systems visible and basing decisions on data instead of gut feeling is what gives us the confidence to change things.
This is less about AI than it sounds. It's about change. And there's more of that coming in our industry whether we like it or not. So what matters is building a culture that can handle change, and ideally one that enjoys it. That's our job as a company, not something to leave to people individually.
That means unlearning. People have years of habits about how you approach a problem, and those habits were right until quite recently. You can't argue anyone out of them. It takes seeing things done differently, often enough that it stops looking like a trick.
And it applies everywhere, not just in development. Everyone has parts of their work worth questioning and processes we keep mainly because we've always had them.
None of that comes from a memo, though. What works is showing what you tried and built, including the parts that went nowhere, until that's a normal thing to do.
That it's only about producing more code. Writing code was never really our problem. Understanding and maintaining complex systems was.
And not just the technical ones. Most of the hard part is working out what someone actually needs, which is rarely what they asked for first. Agreeing on what "done" means. Catching the assumption nobody thought to mention.
Faster code generation alone doesn't get you there. You just build the wrong thing quicker.
A lot of problems don't need AI at all. If something is clearly defined and repeatable, you automate it, with a script, a rule, a workflow. Cheaper, faster, same answer every time. A good share of what people want to point an AI model at is just automation nobody got around to building.
AI excels at fuzzy information, that part is usually at the edges — what comes in and what goes out. Documents, mails, free text, messy data from somewhere else. That's where AI belongs, in the product at least. What happens in between can stay boring.
Put your energy into knowing your domain, building judgment, being able to tell when an answer sounds right but isn't. That's where these tools are weak and where experienced people are strong. They'll give you a confident answer either way, so the things that matter still need checking.
If you don't know where to start, take something you already understand well. Don't try it first on the thing you can't judge. And be careful with the feeling of speed. It's unreliable in both directions, so look at what actually got finished.
I do understand the hesitation. But mostly, do it for yourself. This is your own development, not the company's project. Most of us started out wanting to build great things, and then the years fill up with small obstacles and the great thing never quite gets built. That part is changing. Not all of it, but the "I'd have to learn that whole subsystem first" kind of obstacle is largely gone.
Roles change, they always have. It just feels very fast right now. Which is also the opportunity. These are the rare moments when you get to redefine your own.
Coming from a developer background, it's Claude Code.
What I like is how quickly it gets me into codebases I don't know. I can look around, try things, and have a prototype running that would have cost me weeks before agentic tools existed. That changes what you're even willing to attempt.
It's also what we've rolled out to our developers and the roles around them, so I make a point of keeping my hands dirty with it. It's hard to have an opinion about how people should work if you're not doing the work yourself.