Why multiplication, not automation, is the thing worth chasing.
I've been sat in a lot of conversations recently where the AI questions sound perfectly reasonable on the surface.
How do we save time? How do we get ROI? How do we make sure we're not left behind?
All sensible and commercially justifiable. And almost none of them get to the question that actually tells you something useful.
What is your AI strategy doing to your people?
Not what it's doing for them. What it's doing to them.
I started thinking about this differently a couple of months ago when I was speaking to an SME business. They had a clear AI rollout plan, good exec sponsorship, decent adoption numbers. On paper, it was a success story.
But when I asked what had actually changed about how people worked, they couldn't point to a single role that had become genuinely more capable because of AI. Faster, yes. Lighter, maybe. But more capable? No.
That conversation is what eventually became the AI Maths Framework. And the core idea is pretty simple.
Every AI strategy is doing one of four things to its workforce. It's either dividing, subtracting, adding, or multiplying. The maths tells you what leadership actually believes people are for, even if nobody's said it out loud.
Division
This one's blunt.
AI becomes the reason to cut roles, break teams apart, redistribute work to whoever's left standing. The leadership question, whether it's asked explicitly or not, is: where can we cut?
Division can look sharp in a board pack. Margins tighten. Headcount drops. The narrative sounds decisive.
But the hidden cost is usually bigger than anyone admits.
You lose the institutional knowledge that took years to build. You damage the trust in ways that don't show up in an engagement survey for another 12 months. And you teach the remaining workforce something very specific:
AI is something that happens to them, not something that works with them.
The efficiency story lands well with the board. The cultural damage takes longer to surface, and by then, leadership's already moved on to the next priority.
Worth saying clearly: this isn't transformation. It's cost pressure in a more modern outfit.
Subtraction
Subtraction sounds better than division, and often genuinely is.
Here, AI removes low-value tasks. The admin gets automated, reporting speeds up, first drafts appear in seconds and manual processing shrinks. The leadership question shifts to: what can we take off people's plates?
That's an improvement on division because you're not removing the person.
But subtraction has a ceiling, and I think a lot of organisations will hit it without recognising what's happening.
If you subtract work without redesigning the role around the space you've created, you don't build capability. You just create a gap. And if there's no plan for that gap, it fills with drift, confusion or a kind of low-level anxiety that's hard to pin down.
People start asking questions they probably won't say out loud in a team meeting:
What's my job now, exactly?
If AI can do more of the task, where's my value?
Is this role growing or quietly disappearing?
Subtraction is often where organisations congratulate themselves too early.
Addition
This is where most firms will believe they're winning.
AI tools get rolled out, licences are bought, training sessions happen, prompting workshops appear in the calendar, usage dashboards get reviewed. The leadership question becomes: how do we give people AI?
The intent is usually good. Genuinely. Leaders want to equip their teams, not just reduce them.
But addition has a trap that's easy to miss.
If you add tools without changing workflows, role expectations, decision rights, management habits, or how you measure performance, you haven't transformed the way work happens. You've bolted a new capability onto an old operating model.
I keep coming back to this because it explains where we see AI programmes that create a lot of activity without creating any real advantage. Or the 95% of AI pilots that get disbanded when they hit the real world.
People use the tools, managers mention AI in their updates and the board sees adoption numbers that look healthy. But the organisation is still doing the same work in the same way, just with slightly more assistance.
I'm not sure that's enough. Actually, I'm fairly sure it isn't.
Addition is better than subtraction. But it's not the prize.
Multiplication
This is the level I think is worth chasing. And it's the one I see least often.
Multiplication is where AI doesn't just help people do existing work faster. It helps them do work they couldn't do before.
The leadership question changes completely. Instead of "how do we give people AI?", it becomes "what could our people do that they couldn't do six months ago?"
That one change in perspective lands differently when you sit with it.
- A finance team stops producing reports and starts interpreting patterns, guiding decisions that wouldn't have surfaced otherwise.
- A client team moves from reactive service to proactive insight because they've got bandwidth and data they didn't have before.
- A specialist who was bottlenecked by volume starts operating with genuine range and speed they didn't have before.
In multiplication, the role itself expands. Work becomes qualitatively different. And this is where something interesting happens commercially, because while every competitor in your sector can buy the same licence, far fewer can redesign work, management, governance, and role expectations well enough to actually multiply their people.
That redesign is where the advantage sits. Not in the tool.