The New Frontier

This is a typical interaction with AI:

What can I use AI for?

You use it and you find out it is a lot.

You chat with it, you jump from one idea to another to another, and before long, you find yourself in a state of overwhelm because you realise:

1. You can absolutely do a lot with it
2. You do not have the time to do all that you can think up
3. Your “to-do” list grows exponentially and you don’t do any of it

I wanted to take the time to put down some thoughts I’ve been having when it comes to artificial intelligence, and showcase how I am working alongside AI in this new world we find ourselves in.

The Artificial Intelligence Alongside a Human

My thinking revolves around three components. And these three components come together in a way that unlocks immense benefits.

Firstly, let’s break down how we work into the first two components.

Domain Knowledge & Expertise

This is what you and I have, as professionals in our fields: domain knowledge and expertise.

From our careers and life experiences, we have generated the capability of making judgement and we apply that judgement in many different ways in our work.

This component is the human layer.

Real Use Cases & Data

As we apply judgement, we generate real world use cases and data points.

For example, I applied judgement in making a suggestion to a client that R6.0 is a worthwhile investment over R5.0. This judgement is based on the combination of data, experience, and understanding and it generates many data-points in many different ways. For example, the energy efficiency gain, the relative cost increase and long-term savings, and the calculation of the benefit.

As professionals with domain knowledge and expertise, we have a huge amount of real use cases and data points to call upon.

Coding Capability

This is what was missing and is what changes everything.

Until recently, coding capability came at a significant cost. If you wanted to build a piece of software, you needed to know someone who could code and had spare time, or you painstakingly learnt how to code yourself, or you paid someone else to do it for you. And let’s be honest, you have probably had an app idea or two and considered it, then found the process way too difficult, time consuming, or costly.

Now consider this: having the capability to code is within reach to anyone with a computer.

These three components come together to form a new frontier.

What this might be the new frontier of is totally up to you. It might be the new frontier of business, economics, sustainability, or fashion.

But what is absolutely clear to me is that as someone with domain knowledge and expertise, who has real uses cases and data points available, the capability to code is powerful and changes the very fundamentals of business operation.

The Artificial Intelligence Illusion

The biggest pain point with AI so far for me has been this: you fire up a chat instance with ChatGPT or Claude and ask some questions, and before long, you’ve spent time chatting away but haven’t accomplished anything even if it feels like you have. All you have done is gone around in circles only to feel more overwhelmed than you started.

This is what I call the artificial intelligence illusion. The feeling of progress, but in actual fact, the very real outcome of stagnation.

The answer to this is in taking ownership over this incredible technology.

But what do I mean by “taking ownership”?

Taking Ownership

Artificial intelligence works best when there is persistent memory, guidelines, and clear instructions. It works best when it is yours. From initial start-up, to idea generation, to actual delivery of outputs that you take from A to B.

To open up this new frontier to yourself, the AI needs to be living within your base of operations. Let me explain.

I have moved from the web-based Claude to using Claude Code and have set it up so that Claude Code is an active participant within my local \arkata-hq\ folder. I am writing this article—by myself—using Claude Code as an editing agent next to me as I write.

Claude Code is still using Anthropic’s data-centres via the cloud, but by creating this base of operations over time, I enable myself and my business to one day utilise a local AI that will be fully private and run by local hardware and equipment.

I believe strongly that by taking ownership, and bringing the AI of choice into a folder structure that has been built to enable AI-integration, you will go from the illusion of getting stuff done with AI to actually getting stuff done with AI.

This base of operations is also what makes the third component—coding capability—of the new frontier real. A potential within reach is just that, potential. It is only when we embed the AI into the very substance of our business that things can get done more effectively.

Building Stuff

As you can tell, I’ve gone down the proverbial rabbit hole and am now actually building stuff that makes me more productive and generally very excited about the future of the world around us.

Here is a small, fun example of something I have built. It solved a real pain point I have with Windows specifically. I don’t like that in Windows 11 if I want to switch to light mode or dark mode, I have to dig into the settings four layers deep.

I was in a cafe and this pain point got the better of me and a lightbulb went off above my head as I said: Oh yeah, why don’t I just build a solution to this problem? So, using Claude Code within my computer, I did just that.

Check it out here:

Now this is something small that really anyone could prompt an AI to do.

So what happens when we consider the new frontier thesis? Remember:

In the world of NatHERS, glazing is everything, and as architects, designers, consultants, and builders, we rely on accurate window schedules in the plans and reports.

NatHERS exports messy window schedule CSVs that are difficult to communicate and previously I would—by hand—turn that messy CSV into a usable window schedule. Because the window schedule was built by hand, it took time to create, and because I am human, there is the risk of omission or perhaps doubling up on a window. So, I built a piece of code that bridges the gap between the messy CSV export and a usable, accurate, and replicable window schedule ready for my reports.

Thinking Forward with Artificial Intelligence

I absolutely get the overwhelming feeling because I have been there. But as I learnt more and more and moved from AI-illusion to AI-creation I have become much clearer on how it can actually work to better our businesses and lives.

The window schedule tool started out as me simply automating a manual process. It sits on real data and for the first time, I had the capability to do something about it. When generating the window schedule, there is no AI involved. It is simply code.

This is domain knowledge, real use cases, and the ability to code converging to form a new frontier ready for me to explore.

So here’s what I’d ask.

If this article resonates, if you’ve felt that same overwhelm, or you can see your own domain knowledge and data sitting there waiting for that third leg, comment or direct message me! I’d love to hear how you’re thinking about AI in your business.

Until next time,

Ben

P.S. I intentionally put aside the environmental questions we all have over AI’s sustainability and instead will address that in its own future edition of Thinking Forward.