When you start an AI company there’s one thing no one tells you about: that you can't just have the vision; you have to build it yourself first. Not because you're a control freak or because you don't trust other people, but because the thing you're trying to create doesn't exist yet. And if you can't show people what "it" looks like, you're just another person with a PowerPoint deck full of aspirational buzzwords.

I've been called a CEO in the code. And yeah, I love that. Because for the first chapter of Agentiiv, that's exactly what I was – hands on keyboard, writing prompts, testing agents, breaking things, rebuilding them, and slowly figuring out how to make humans and AI actually work together in ways that felt useful instead of gimmicky.

But the real value of being in the code isn't about staying there forever. It's about knowing when to build, when to bridge, and when to get the hell out of the way.

Bringing Two Worlds Together

Let me back up a bit. My background is running large teams – we're talking 200 people, $200 million budgets, cross-functional chaos across strategy, marketing, finance, HR, and legal. When you operate at that scale, you develop what I call systems thinking. You start seeing the invisible architecture of how organizations actually work - the workflows, the handoffs, the bottlenecks, the moments where everything either flows beautifully or falls apart completely.

Then came November 2022, and suddenly everyone had access to GPT-3. A model that could barely make bad rhymes, and yet it felt like magic.

I saw something nobody else seemed to see. The intersection between decades of organizational systems expertise and this new capability called prompt engineering could create something fundamentally different. Not chatbots. Not generic AI assistants. But what we call digital twins – domain experts designed around complex human workflows and expertise, powered by AI but shaped by deep understanding of how people actually work.

So I started building. Not because I'm a developer by training, but because I needed to prove the concept to myself first. Could you actually design AI agents around humans instead of forcing humans to adapt to AI? Could you create systems that augmented expertise instead of replacing it?

Turns out, yes. But only if you understand both sides of the equation.

Building in Shifting Sand

AI is different from every other technology wave because the foundation keeps changing underneath you.

We're living through what I call the 1,000 days of AI. In November 2022, GPT-3 could write mediocre marketing copy and make those aforementioned bad rhymes. Today? Gemini 3 is fully multimodal with advanced reasoning capabilities. It can create sound, image, and video from a basic prompt. It's starting to understand the physics of the world.

Think about that evolution. What you purpose-built for older models won't work the way it's meant to with newer technology. The systems you created six months ago might already be obsolete. The workflows you optimised for one generation of AI need to be completely reimagined for the next.

You're simultaneously dealing with two transformations happening at once, and it feels impossible sometimes.

The AI itself is evolving daily. New capabilities, new models, new architectures, new possibilities that didn't exist last week. Meanwhile, human workflows are changing as people learn to work with AI. The way your team collaborates with AI agents today is different from how they collaborated three months ago. They're learning, adapting, discovering new ways to leverage these tools that you couldn't have predicted.

So the CEO who's building in agentic systems needs to bridge both of these moving targets. You need to understand how the humans are changing and how their workflows are evolving as they work more with AI. You need to understand how the AI itself is morphing and transforming on a daily basis.

You can't do that from a spreadsheet. You have to be in it.

Learning When to Let Go

But, and this is the part that took me a while to accept, you also can't do it forever.

As the technology has advanced, the infrastructure and scaffolding that makes it all run has become exponentially more complicated. What started as prompt engineering has evolved into graph RAG, vector databases, complex multi-agent orchestration, and technical considerations that require specialists with machine learning master's degrees.

I had to face a hard truth. To realise the full potential of what we're building at Agentiiv, we needed engineering sophistication that was beyond my expertise. And more importantly, my time was better spent elsewhere.

My value isn't in understanding every technical detail of our infrastructure anymore. My value is being the bridge between the human side and the technical side. Understanding what users need, how their workflows are evolving, where the pain points are, and translating that into product vision that our engineering team can build brilliantly.

That's where the CEO-CTO partnership becomes critical. Michael Carrick, our CTO, is in the weeds every day. He knows how to build sophisticated agentic systems. He understands the technical architecture at a level I never will. But I bring the user point of view, the organizational systems thinking, the understanding of how humans need these tools to work.

And honestly? Even Mike can't go as deep as we need anymore. That's why we've hired specialists – machine learning engineers, cloud architects, infrastructure experts who can dive into the complexity while Mike orchestrates the vision.

It's like sending your kid to school. Sure, you could try to teach them physics yourself, but they'll learn it better from someone who's devoted their life to teaching physics. Your job is to make sure they're going to a good school and developing into a well-rounded human. Not to teach every subject yourself.

Lessons for Leaders in Technical Spaces

What I've learned about being a CEO in technical spaces, especially in AI where everything is moving at lightspeed, comes down to a few key principles.

Get your hands dirty first. You can't have a vision for something you don't understand. Build it yourself until you can articulate exactly what you're trying to create and why it matters. Your team needs to see that you understand the problem at a visceral level.

Stay technical enough to bridge. You need to speak both languages - the human workflow language and the technical architecture language. You don't need to write production code, but you need to understand what's possible, what's hard, and where the technology is going.

Know when you're the bottleneck. When staying in the weeds means you can't focus on vision, strategy, and understanding users at the level that only you can bring, that's when you need to step back. Your role shifts from builder to orchestrator.

Hire people smarter than you. This sounds obvious, but it's hard in practice. Bringing in specialists who know more than you do about specific domains means letting go of control and trusting their expertise. Do it anyway.

Retain the bridge role. Even as you step back from building, you can't completely disconnect from the technical side. You're the translator between what users need and what engineers can build. That's your ongoing value.

Building for the Future at Agentiiv

This philosophy isn't just how I've approached leadership. It's embedded in how we build our platform.

Agentiiv exists because we understand both sides of the equation. We know how humans work because we've lived it. We know how AI works because we've built it. And we know that the magic happens at the intersection - when AI agents are designed around human workflows instead of the other way around.

Our digital twins – whether they're research specialists, UX analysts, or executive coaches—aren't just fancy chatbots. They're domain experts designed with deep understanding of how professionals actually do their jobs, powered by cutting-edge AI, and architected by engineers who understand both the human side and the technical complexity required to make it work at enterprise scale.

We're building tools that augment human expertise rather than replacing it. Because I've been on both sides – as an executive managing complex organizations and as someone in the code building the actual agents – I know what's possible when you bridge those worlds correctly.

The Evolution of Leadership

We're at a unique moment in history. AI is transforming every industry, every workflow, every job. The leaders who will succeed aren't the ones who stay purely strategic or purely technical. They're the ones who can move fluidly between both worlds.

You need to understand the technology deeply enough to spot opportunities and bullshit. You need to understand human workflows intimately enough to know where AI creates real value versus where it's just noise. And you need to know when your job shifts from doing to orchestrating.

That's the CEO in the code - someone who builds until they can articulate the vision, who stays technical enough to bridge two worlds, and who knows when to step back and let specialists take it further than you ever could alone.

Because the thing you're building will outgrow you. And if you've done your job right, that's exactly what should happen.

The question is whether you'll recognise that moment and evolve with it, or whether you'll hold on too tight and become the bottleneck to your own vision.

I chose to evolve. And Agentiiv is stronger because of it.