I gave a talk last week to a City CIO Club gathering on the next generation of skills leadership teams need to develop around AI.
Afterwards, one of the CIOs in the room stood up:
"We are on this right now, but only in the tech department. And actually mostly me. I am axing tens of millions of pounds of spend because we can now use AI to build greater capabilities for far less. My team hates me for it because their jobs are at threat.
Security platforms we have paid for for years, workflow tools and analytics software. The vendors can't believe how cheeky I've become."
Then he paused.
"But I don't think many people yet understand how big the impact on their business is going to be. Or for CEOs, how big it should be."
He is right. There are now two types of C-level leader.
Those who have an AI strategy, meaning they are using it for serious leverage, taking millions of pounds of cost out of the business and redeploying that capital. Their first question is:
"How many people can I fire?"
And those who would rather see how it goes. The second group believes the market will move slowly enough for them to catch up.
It will not.
Vendors and investors are spooked
The products being scrutinised are deeply embedded software that have run large operations over the past decade or more.
The enterprise software giants are still posting revenue growth, but the numbers are a mirage.
Their contracts are long and switching is painful, so the income reflects decisions made one to two years ago.
Underneath, the cracks are already showing. Seat counts are falling as companies become more efficient and net revenue retention has stalled.
Salesforce and ServiceNow have both lost about a quarter of their value this year. HubSpot has fallen 39%, Atlassian is down 35% and Adobe has taken a beating because investors are pricing in disruption.
DIY for now
The market is repricing an entire category. Where software deals once felt impossible to do without, they are no longer permanent because AI is giving buyers leverage to do it themselves.
They are becoming a commodity. Perhaps not yet entirely, but watch how power shifts to the buyer.
In-house teams are discovering they can replicate significant portions of what these platforms do for a fraction of the licence cost. The vendor sales teams arrive for the renewal conversation expecting the familiar ritual - add 10-20% and walk away.
In a growing number of cases, they find the decision has already been made.
"I've been demanding 75% of the cost is removed for us to stay with the vendor or we will find another way," another CIO said.
Head in the sand
I have the privilege of talking to a large number of CEOs and senior leaders.
Some have barely used AI for anything more than a search query. Others are building entire applications and internal workflows themselves with no technical background at all.
The gap between those two groups is widening every month and most organisations have not yet come to grips with four things.
- What they could use AI for today if they deployed it at full capacity.
- What the new version of full capacity looks like for each person in the business.
- What skills, technologies and vendor relationships they no longer need.
- Perhaps most importantly, what they could achieve if they redirected their attention toward new capabilities rather than defending the old ones.
Sourcing technology to turn strategy into reality
Proudly sponsors the Executive Summary
Speed to redeploy
The redeployment of people vs AI question comes up in almost every conversation. That is, if we restructure, can we move the people into new roles?
The answer has two parts. If those people have not developed AI fluency, there is very little chance they will be fit for the roles the new structure requires.
And even if they were capable of retraining, the pace at which the required skillset continues to evolve means they would likely arrive behind the curve. The retraining programme cannot run faster than the change it is trying to catch.
So only the superkeen will make that leap.
The orchestrator: a new type of operator
One huge gap in the market is a definition of what skillset and service delivery models are going to be.
Where this was once a question that could take consultancies months to find out, it won't be such a biggie for companies because they can find out much faster now.
Using AI, they can launch 20 campaigns or propositions simultaneously and test until they find the right answer, using the market help them lead.
If you think I am exaggerating, look at JP Morgan, which used AI to generate and test hundreds of variations of marketing copy simultaneously across personal banking, home lending and wealth management.
The result was click-through rates up to 450% higher than human-written ads.
The definition of skillsets then needs to be specific, because in boardroom conversations, words like transformation, augmentation and future-proofing, describes a direction without describing a destination.
The clearest model I have found starts with a single image: the solopreneur on steroids. Someone who can route traffic, free up friction, build, communicate and deliver, hunt and drop, with AI handling execution across all simultaneously.
They are working a different scale entirely with AI agents doing a lot of work for them. Productivity lift is huge.
Where CEOs are getting it wrong
A lot of CEOs are looking at this through one lens: how many people can I fire?
The bigger question is how do we nuke our competitors using productivity and speed they cannot match yet.
One question is reactionary, the other is visionary.
Where do you sit?