I get asked a version of this question a lot: "This is clever, but couldn't I just get Claude to do it myself?"

It's a fair question. And because I build an AI-powered platform for a living, you might expect me to bat it away. I won't. The honest answer is: yes, you can get AI to do some of this yourself. But there's a clear line where do-it-yourself stops being smart, and it's worth understanding exactly where that line sits before you bet your execution on it.

First, a story about getting it wrong

Let me start with a confession. In a previous venture, we spent the best part of a year building an AI sales-coaching product. We designed it, built it, launched it and then it got taken out at the knees almost overnight when Claude arrived and did much of the same thing out of the box. A year of work, overtaken by something that barely existed when we began.

So I'm not romantic about this. I know first-hand how fast general-purpose AI is moving, and how much it can do. If anything, that experience made me more respectful of what a sophisticated operator can now build for themselves with the right prompts.

Which is exactly why I want to be straight about where it helps and where it doesn't.

What AI can genuinely do for you today

If you're hands-on and you know your way around these tools, you can absolutely use AI to do useful, discrete pieces of work. Draft a goal framework. Summarise a strategy document. Interrogate a single dataset. Pull together a one-off analysis for a board paper.

For a specific person doing a specific task, that's real value, and you should use it.

The problem is that strategy execution isn't a single task. It's a continuous, connected system that touches every person in the business, every week.

Where do-it-yourself falls down

Here's the line: General-purpose AI is a point solution. Brilliant at one task, for one person, in one moment. Strategy execution is the opposite of that. It needs to run across the whole organisation, pull live data from the tools you already use, keep it connected to your strategy, and keep doing it reliably as things change.

To replicate that yourself, you'd be asking your team to wire a general AI tool into your CRM, your project management system, your HR platform and your financial data and then maintain those connections, manage the security, and keep it all coherent. That's not a prompt. That's a build.

And it's a build with a real opportunity cost. If your engineers are great engineers, you want them shipping the product that wins you the market, not spending their weeks plumbing Claude into HubSpot and Asana to recreate an execution layer. The few years we've spent doing exactly that hard work is the whole point: so you don't have to.

The part most people underestimate: your data

There's a second reason I'd think carefully before routing your most sensitive operational data through a general AI setup, and it's not a small one.

When you're feeding in performance information, strategy documents and live business data, where that data goes matters enormously. We spend a great deal of time making sure customer data stays within your organisation, data that it isn't quietly shipped off somewhere you don't control. On top of that sits a growing weight of regulation, from data-privacy obligations to the EU AI Act, that governs how these systems can use your information.

That's a lot of finer detail and real risk to manage on your own, and it's easy to overlook in the excitement of "look what I got it to do." A platform built for this has already addressed it as a condition of existing.

AI should be the engine, not the headline

The mistake I see is treating AI as a magic trick, as if the technology itself is the answer. It isn't. The market is full of sophisticated buyers who, quite reasonably, want to know how a tool reaches its conclusions before they'll trust it in front of a board. "It's AI" is not a reason to believe something.

So the way I think about it, and the way we built RoleKick, is that AI is the engine, not the message. It does the heavy lifting: cascading strategy into goals, pulling and interpreting data from across your systems, surfacing where execution is drifting. But a human reviews and approves before anything is published, you can see the working rather than just the output, and your data stays where it should.

So, can you just use AI?

For a one-off task, yes - go and do it. For running strategy execution across a scaling business, reliably, securely, connected to everything else? That's a system, not a prompt, and building it yourself is rarely the best use of your best people.

The deeper truth underneath the question is that doing nothing is no longer the safe option either. If your systems are two or three years old, you're already behind what's now possible, you’re carrying the risk and feeling none of the benefit. The pace isn't slowing down.

The opportunity is to get the competitive edge these tools offer without betting a year of your engineering team on rebuilding it from scratch.

Want to see what a purpose-built execution layer does that a DIY setup can't? Book a Live Strategy Build and we'll show you on your own business.