Most of Your Leaders Aren't Using AI. That's Now a Business Risk.

Adam Cogan is the founder of SSW, a software development and AI firm with over 30 years of enterprise experience. He sat down with Dave Kenyon from OAHI's Evolving Workforce podcast.

I just did a keynote at a conference of about 50 business owners. All running companies between $30 million and $100 million in revenue. Successful people. Leaders who know their industries cold.

I asked them: "Who here has a team of more than five software developers?" Nothing. "More than two?" Nothing. "One full-time developer?" Not a single hand.

These were not small operators. They were bigger businesses than mine. And they had no one building or running technology for them. That part I expected. What I did not expect was how many of them were not using AI either, not really. They knew it existed. Some had dabbled. But as decision-makers, they had not actually gotten their hands on the tools. That gap between knowing AI exists and understanding what it can do is now a competitive liability.

The problem is not the technology. It is the distance from it.

I have spent thirty years running software teams. For the last few, I have been running software teams that are massively powered by AI. And the pattern I keep seeing in Australian organisations, especially mid-to-large ones, is that the leaders furthest from the tools have the most distorted view of what those tools can do.

They either think AI can do everything and it is terrifying, or they think it cannot really do much for their specific situation. Both views are wrong. And both come from the same root cause: they have not used it themselves.

When you are running a company with hundreds or thousands of employees, you have people to do things. Minions, as I sometimes say. That insulation means you can go a long time without directly encountering the technology that is reshaping how work gets done. But the cost of that distance is growing.

What AI is actually changing right now

Here is where I see genuine, measurable value, not hype.

Coding tasks, reconciliation, document generation, anything to do with spreadsheets. These are the areas where AI delivers a step-change, not an incremental improvement.

I will give you a concrete example. Our system administrators, not developers, had been collecting a bunch of logs tracking Microsoft partnership activity across clients. They sent me a spreadsheet with multiple tabs. I was supposed to understand it before a meeting. I used the Claude Excel add-in, asked it to add a graph and summarise what was going on. Vague prompt. It gave me a beautiful graph. I moved it to the first sheet, asked it to apply our company colours, and walked into that meeting looking like I had spent an hour on it. I had not.

That is not a developer workflow. That is a business leader workflow. The tools have crossed that line.

The areas that are still mostly hype are AI-generated images and video production, useful and semi-useful, but not the step-change you get from analytical and document tasks. And anything labelled "AI-powered" that is just autocomplete wearing a suit.

The AI slop problem nobody is talking about honestly

There is something I call AI slop, and it is doing real damage to organisations that think they are adopting AI well.

A mate of mine, not a tech person, showed me a message he had received from a work associate. He was confused by it. Was it passive aggressive? Why did it have all that extra padding in it? I read it immediately and knew it came straight out of ChatGPT. The rule of threes. The flowery language. The emojis nobody actually uses in business communication.

He could not tell. But something felt wrong. And that is the problem with AI slop. It erodes the trust that makes communication work.

This is not just a LinkedIn post problem. It is showing up in internal communications, in client emails, in reports that go to boards. When people feel they have received something inauthentic, something generated rather than considered, it registers. Even if they cannot name it.

The fix is simple but most people skip it. Write your genuine email first. That is your real message. Then do not ask AI to "reword this to be friendly" because that is where slop gets injected. Instead ask it: "Please review this and give me changes numbered in the form of change X to Y." Then take only the suggestions that actually improve it. Your voice stays intact. The communication stays real.

What good AI adoption actually looks like inside an organisation

The first thing I did at SSW was get AI into the hands of every single employee. Not just the developers. The TV production team. Marketing. The accountants. Anyone not using it needs coaching, and if they will not, that is a conversation about whether they are the right fit for where the company is going.

We put AI use into our induction process. From day one, new staff know this is how we work. We use it transparently. When they complete training modules and get answers wrong, they work through those gaps with AI before speaking to a person. Same principle with client emails: do not send one without running it past an AI tool first and asking for numbered feedback.

The other thing I would strongly recommend is appointing a VP of AI. Someone whose job it is to stay current, share what they are learning, and pull the organisation up to a baseline. You need a person that others know to go to. In time, that role becomes C-suite, a Chief AI Officer. I have not seen many yet, but they are coming.

The companies I work with that are doing this well have one more habit worth stealing. In every team review, each person shares what AI tools they used that fortnight, what worked and what did not. That knowledge-sharing is how you compound the gains. Otherwise you have pockets of competence and large swathes of the organisation still working with a chisel when everyone else has a chainsaw.

The workforce management angle that most leaders miss

I was at the Australian Payroll Association Summit recently and it struck me how much of the work in payroll, rostering, and workforce compliance is exactly the kind of busy work that AI is built to eliminate.

Reconciliation tasks. Award interpretation. Checking whether a timesheet is compliant with a Modern Award or Enterprise Agreement. These are repetitive, rules-based, high-stakes tasks. The risk of getting them wrong in Australia, under the Fair Work Act with wage theft now a criminal offence, is serious. The manual effort required to get them right is enormous.

Organisations like OAHI are already applying this logic to workforce management, automating award interpretation, the Better Off Overall Test, and the payroll compliance checks that used to require a specialist sitting down with each pay run. The technology exists. The regulatory pressure to use it is only increasing.

But none of that technology delivers value if your leaders do not understand what it does or why it matters. That is the gap I keep coming back to.

Start here

If you are a business leader in Australia and you are not personally using AI tools yet, that is your first task. Not delegating it. Not reading about it. Using it.

Try Claude Code. It is not just for developers. You can ask it to build you a webpage that summarises key stats, generate a report, or reorganise a spreadsheet. The point is not the specific output. The point is that you build an accurate model of what the technology can and cannot do. Right now, a lot of leaders are making decisions about AI strategy based on secondhand information, and that is a risk your competitors are not carrying.

The organisations that will be in a strong position in two to three years are the ones where AI adoption started at the top, not the bottom.

Adam Cogan is the founder of SSW, an enterprise software development and AI firm based in Australia. This piece draws on his conversation with Dave Kenyon on the Evolving Workforce podcast.

Check OAHI Pay Pulse for real-time BOOT compliance and underpayment prevention.

‍

‍