The New AI Role

The Role Your Business Doesn't Know It Needs

By David Rogers

Something is shifting in how businesses build software. Quietly, but quickly.

For years, the default was simple: need a custom tool? Hire an agency. Need a dashboard, an intranet, an internal app? Write a brief, sign a contract, wait three months, pay the invoice. The agency owns the knowledge. You own the output. Until something needs changing, then you call the agency again.

That model made sense when building software was genuinely hard. It's becoming less hard. Not easy, but meaningfully more accessible than it was two years ago.

AI hasn't just changed what software can do. It's changed who can build it.

The Prototype That Never Leaves the Room

Here's what I'm seeing happen inside organisations right now.

Someone gets handed a mandate. Use AI to save time, find efficiencies, build something useful. They're smart. They're motivated. They open a tool, start prompting, and build something that genuinely impresses people in the room. A report that used to take a day now takes ten minutes. A process that lived in someone's head is now documented and automated.

Then they try to share it.

The HTML file doesn't work on someone else's machine. The data it relied on is in a spreadsheet only one person can access. The whole thing is held together with good intentions and a browser tab. What looked like a solution was a prototype that never had a chance of surviving contact with the rest of the business.

I've seen this play out more times than I can count. The problem isn't the idea. The problem isn't the person. The problem is that building something real, something stable, secure, and usable by more than one person, requires a different set of skills than building something that works on a Tuesday afternoon.

What Businesses Actually Need

The capability gap isn't about AI literacy. Most forward-thinking organisations have people who can prompt their way to something useful.

What they're missing is someone who can take that useful thing and turn it into a product. A corporate dashboard that the whole leadership team can log into. A custom intranet that reflects how the business actually operates. An internal app that replaces a manual process and keeps working six months later when the person who built it is on leave.

These aren't automations. They're software products, and building them requires someone who understands the full distance between a good idea and a deployed, maintained, working tool.

That person needs to know where data should live. How to make something accessible without making it insecure. When to build and when to buy. How to talk to an executive about what's possible and to a developer about how to do it.

This is the AI Product Architect and most businesses don't have one.

Replacing the Agency Model

Here's the part that changes the economics entirely.

Historically, a custom internal tool meant an agency engagement. Months of back and forth, a brief that never quite captured the vision, a handover that left the business dependent on external support for every future change. Not because agencies aren't good, many are, but because the model creates dependency by design.

The AI Product Architect changes this. With the right person in-house, businesses can conceive, build, and own their own tools. Not prototypes. Not demos. Real software, built to their specifications, maintained by someone who understands the business from the inside.

The barrier to building has dropped. The barrier to knowing what to build, and how to build it properly, has not.

That's the distinction. AI gives you the capability. The AI Product Architect gives you someone who knows how to use it.

What This Role Looks Like in Practice

This isn't a developer. It isn't a prompt engineer. It isn't an IT manager who's good with tools.

It's someone who sits at the intersection of business understanding, technical capability, and product thinking. When you're looking for this person, whether hiring or engaging, here's what actually matters.

Can they ship something real?
Not a proof of concept but a working product that other people can use without hand-holding.

Do they think about security from the start, not as an afterthought?
Where does the data live, who has access, what happens if something breaks.

Can they scope honestly?
The right person will tell you what's realistic, not just what's exciting.

Do they understand your business, not just the technology?
The best tools are built by people who understand the problem deeply enough to question the brief.

Can they own it long term?
Building it is one thing. Being accountable for it is another.

Where This Is Heading

The AI Product Architect is the title that makes sense today. It's legible, it's hireable, it maps to something executives already understand.

But the leading edge of this role is already moving somewhere more interesting. The best practitioners aren't waiting to be handed a brief. They're working at the inception stage, before the problem is fully articulated, asking what an AI-native business could build if it started from scratch instead of automating what already exists.

That's a rarer skill. And it's where the real competitive advantage will come from in the next few years, not in the businesses that use AI, but in the businesses that build with it.

Most organisations will hire for this role once it has a standard job title and a flooded talent market. The ones that move now will be ahead of both.

I help leaders understand this capability, define what they need, identify the right people, and in some cases build it myself. If this is a gap you're staring at, get in touch.