Paul Deraval, Cofounder & CEO of NinjaCat, is a software veteran with 20+ years driving innovation in martech, AI and agency growth.
Building and experimenting with AI has changed how I think in ways I never expected. Not simply how I approach technology issues, but regarding work, leadership and what it actually means to solve a problem.
At first, like most people, I treated AI as a tool. Something to layer into existing workflows to make them faster, cheaper or more convenient. But that framing breaks down quickly. The moment you start working closely with AI agents, you realize that the real shift isn’t what the technology can do, but how it forces you to rethink your approach to problems in the first place.
Here are five ways AI has fundamentally changed my thought process.
Start With End State, Not Process
I used to approach problems by asking: How do we improve or automate what already exists? That instinct doesn’t hold up anymore. Now, I start with a different question: What’s the most ambitious version of this outcome I can imagine? Not how we do it today, but what “done right” actually looks like.
There was a weekend when I made over 75 commits to a side project, not grinding away at my desk, but delegating work to AI between gym sets and cleaning out my garage. I wasn’t thinking about the steps. I was thinking about the outcome, and then letting the machine handle the path. That shift alone changes everything. You stop optimizing workflows and start redefining them.
Learn To Prove, Not Just Say
One of the most unexpected changes has been how I lead. There was a time when you could simply tell your people about an idea for the business and get buy-in. That doesn’t work anymore. The gap between what AI can do and what people believe it can do is too wide.
So now, I build. I spend a significant amount of time creating proofs-of-concept with the goal of helping my team understand the possibilities, not to simply create the first iteration of a new tool. I’ve gone from “I think this is possible” to “I know this is possible” to “Let me show you.” That’s become a core leadership skill: making the future tangible by showing it to be true, not simply knowing.
Assume Until Proven Otherwise
I used to default to skepticism. If something sounded too complex or too abstract, my instinct was to assume AI couldn’t handle it. Now I operate in reverse.
When a problem surfaces, my default reaction is: it probably can do that, let’s try. Sounds like a small shift, but it’s powerful. It opens up experimentation. It lowers the barrier to testing ideas. It removes the mental friction that stops progress before it starts. I tell my team the same thing: assume it can, and ask for help figuring out how. Naivety, in this moment, can be a competitive advantage.
New ROI Conception, Cost Of Standing Still
One of the more subtle changes in my thinking is how I evaluate impact. We’re used to measuring ROI as something positive, using words like growth, efficiency and output. But with AI, there’s another dimension: the cost of not adopting it.
There are investments you make that don’t show up as a gain, but would absolutely show up as a loss if you didn’t make them. Staying at the bar is now part of the equation. Not falling behind is a form of return. That’s a different way of thinking about value, and it changes how you prioritize.
Different Resistance, Structural Not Personal
AI adoption isn’t just a technical challenge; it’s also a human one. I’ve spent significant time thinking about why people hesitate to engage with AI, and I’ve stopped viewing it as stubbornness or lack of effort. It’s something more predictable.
There’s a progression: comfort zone, fear zone, learning zone, growth zone. Most people hit that fear zone and interpret it as a signal to stop. But it’s actually a signal that they’re exactly where they need to be. Understanding that has changed how I lead. It’s not about pushing harder, but rather about helping people move through that transition.
Where AI Agents Changed Thinking Most
If there’s one area that accelerated these shifts, it’s working with AI agents. They force you to think in terms of context, not just capability. If you built the agent, you should know what it’s capable of, so the challenge becomes understanding whether or not it has the right information, constraints and environment to act effectively.
We built an internal troubleshooting agent to help our engineering team diagnose issues. The first version looked impressive. It could suggest fixes, identify patterns and even resolve some problems end-to-end. But it was inconsistent. Sometimes right. Sometimes confidently wrong. The agent lacked the context to do its job effectively.
Our engineers carry years of implicit knowledge: edge cases, historical quirks, things that don’t exist in documentation. Once they started actively “training” the agent like a normal employee by feeding it that context, refining its access and pressure-testing its output, it changed. It became less like a demo and more like a junior teammate.
That experience reshaped how I think about systems entirely. Not as platforms or tools, but as systems of context and environments where agents can operate effectively.
AI hasn’t just made me faster. It’s made me think differently. It’s pushed me to start with outcomes instead of processes. To prove instead of assert. To assume possibility instead of limitation. To rethink value, and to better understand how people adapt to change. But most importantly, it’s clarified something simple: the real advantage isn’t simply access to tech, but how you use it. AI may start as a tool, but if you nurture it and let it grow alongside your team, it becomes something else entirely: a forcing function for better thinking.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
