Ignore the hype. Pivoting to an agentic-first business isn't a piece of cake (2026)

You’ve heard the hype about agentic AI. They say it’s going to completely change the way we work.

Maybe you’ve seen some scepticism too. Like Gartner’s prediction that more than 40% of agentic AI projects will be cancelled by the end of 2027 due to escalating costs, unclear business value, and inadequate risk controls.

But have you heard from a real business rebuilding their own operations to be agentic-first right now?

That’s exactly what we’re doing at my mid-sized company, Enate. And here’s an honest look at what it looks like from the inside. Plus, the mistakes we’ve made along the way.

Our vision

At Enate, we want to reach the point where we’re able to build software without anybody writing code by hand. We’ll use agents to run a software product factory, then apply that same factory thinking to the rest of the business.

That's not a project we can bolt onto how we already work. If agents just sit on top of the old structure, you inherit all its friction and add a new layer of it. So we're restructuring how our operations run.

While the aim is zero coding, we’ll still be shipping enterprise-grade products. It’s just that after there’s no technical barrier to software development, everyone in the business will be able to design their own updates and improvements to our software.

I believe this is the future of software development. An agile team able to ship new features without having to wait for the dev backlog to be cleared. We’re just getting ahead of it.

Agentic-first doesn't mean slop

Now I know what you’re thinking. “Nobody writes the code” means trading quality for speed.

But for me, going agentic-first is about being more deliberate about craft, not less. I’m an engineer by trade, and this is where the manufacturing mindset comes in. It’s all about obsessing over the machine that builds the output, so the output is consistent and good every time.

I've written before about why AI slop is a race to the bottom. That’s not the route we’re going at Enate, and it isn’t the route I'd recommend you go down either. Care and human judgement are always what are going to carry value in service businesses. But going agentic-first is how to protect that moat at scale in the age of AI.

The real challenges

So that's our ambition. But the truth is that it's hard. We’re changing how people work, and not everyone finds the shift easy.

The biggest roadblock we’ve hit is trying to get software engineers to adopt the mindset of manufacturing engineers. We still see them as craftspeople, but we need them to become craftspeople who think about the machine, not the part. So when an agent gives them buggy code, we don’t want them debugging the code. We need their first instinct to be to improve the machine that wrote that code.

We’re building the plane while flying it here, which means we’re still figuring out how to help our people through this change. The best advice I can give to anyone starting this journey is to put your energy into helping people think like manufacturing engineers, not debugging their code.

Then there's the orchestration problem. Your product factory needs to run against the dev ops pipeline that governs how software actually gets shipped. But you also need an agent sitting above that, deciding what gets built next and driving the build against a specific user story. Get the handoff wrong and you’ll end up with agents confidently shipping features nobody asked for.

And a major headache at the moment is consistency. When everyone in the business is vibe coding with their own agent, your style guide goes out the window. How do you keep UX design consistent across everything the factory produces? That’s an open problem we're trying to solve right now.

Over to you

So where do you actually start?

Audit your business like a manufacturing system, then rebuild it from the ground up. For us that means three things: A system of record (eg. HubSpot), systems of action (the agents doing the work), and an orchestrator that coordinates it all.

None of that is going to come with a manual, and you won't be able to plan it all up front. So instead, think like a startup. Decide roughly where you're headed, start building, and course correct as you learn what doesn't work.

Rebuilding around agents this way is a huge departure from how we’re all used to working. So expect resistance from your team and do all you can to manage them through that change.

Lastly, don’t forget, there's an early mover advantage here, and it won't last. In two years everyone will be running their business - at least in part - agentically. Start thinking agentically now to get ahead.