Last month, I watched something that should terrify every founder running a company built before 2023.
A colleague, running a project we are working on in stealth, built a software product in a few weeks.
One amazingly talented person. A team of AI agents. Done.
2 years ago, that same product would have taken three to four months and £200,000+ to build. A team of engineers. A product manager. A designer. QA. Project management overhead. The whole machine.
Now: one person and agents.
Staggering.
And here’s the part most CEOs are still missing: this isn’t the future. It already happened. I watched it. I’ve watched it happen multiple times now, across multiple companies, in multiple sectors.
If you’re running a company that was founded before AI got serious, you need to sit with something deeply uncomfortable:
The world your company was designed for no longer exists.
The org chart you built. The roadmap you committed to. The cost base you optimised. The hiring playbook you ran. The way you make decisions, ship product, serve customers, structure teams. All of it was designed for a reality that quietly ended in the last 18 months.
The companies that survive the next 24 months won’t be the ones that “added AI.”
They’ll be the ones that rebuilt around the world that actually exists now.
This essay is about what that actually means. Not in theory. In practice. From inside a company that’s living through it.
The pattern I’ve seen before
I’m 36. I’ve been building tech companies for 15 years. In that time, I’ve watched four major shifts:
The mobile shift (2010–2013)
The cloud and SaaS shift (2012–2016)
The remote work shift (2020–2022)
And now, the AI shift (2023–onwards)
Each one followed the same pattern.
A small handful of companies, usually founder-led, usually willing to make uncomfortable decisions, rebuilt their operating model around the new reality.
The rest did the comfortable thing. They formed a committee. They added the new thing as a feature. They hired a “Head of [insert shift here].” They sent a memo. They declared victory.
And then, quietly, over 24–36 months, they lost a decade of compounding to the companies that did the harder thing.
I’ve seen this pattern play out so many times now that I can predict it with depressing accuracy.
What’s different about AI is the speed and the magnitude.
Mobile took five years to fully reshape the consumer landscape. Cloud took roughly the same. AI is happening in 18 months. And the leverage gap between augmented and non-augmented teams isn’t 2x.
It’s 100x. I actually think its more.
Which means the cost of running a company designed for the old world isn’t a competitive disadvantage. It’s an extinction event.
The lie you’re telling yourself
Most pre-AI companies are doing the same thing right now. I see it constantly. I’ve sat in dozens of board meetings where a perfectly intelligent leadership team has convinced themselves they’re “doing AI.”
Here’s what “doing AI” usually looks like:
They’ve stood up an AI committee.
They’ve bought ChatGPT Enterprise or Claude for Work.
They’ve added “AI-powered” to a product page.
They’ve sent a memo about “exploring use cases.”
They’ve added a chatbot to their website.
They’ve automated one workflow nobody really cared about.
That isn’t transformation. That’s theatre.
It looks like progress because there’s motion. There’s activity. There’s something to put on a slide. But the underlying business, the org chart, the cost structure, the product roadmap, the way decisions get made, the kind of people being hired, is functionally identical to what it was in 2022.
You’re running a 2022 company in a 2026 world.
And while you’re doing that, three things are happening:
A new generation of AI-native competitors is being born with a fraction of your headcount, a fraction of your costs, and a roadmap that ships in weeks instead of quarters.
Your best people are looking around and realising the company isn’t serious. The really capable ones, the ones who could be 10x with the right tools and mandate, are quietly leaving for companies that are.
Your customers are using AI themselves. They’re getting better at evaluating what’s possible. The bar for what counts as “good” is being reset every quarter, and your product is sliding down it.
You don’t notice until it’s too late. Because companies designed for the old world usually look fine on the surface. Revenue is steady. Customers are renewing. The team is busy.
But the foundation is rotting.
What rebuilding actually means
When I say rebuild, I don’t mean “use ChatGPT more.”
I don’t mean adding an AI feature to your product.
I don’t mean training your team on prompt engineering.
I mean a genuine, structural rebuild of how your company operates. The kind of rebuild that makes people inside the business deeply uncomfortable.
Here’s what that looks like in practice. Pick the ones that apply to you. Most of them probably do.
1. Rebuild your org chart from zero
This is the one nobody wants to do. But it’s the most important.
Don’t ask: “How can AI help our current team do their jobs better?”
Ask: “If I were starting this company today, with the tools available now, what would the org chart look like?”
You will arrive at a much smaller, much more senior, much more leveraged team. Almost guaranteed.
The companies winning right now are running with org charts that would have looked impossible two years ago. One operator doing what used to require five. A team of three building what used to need a department.
This doesn’t mean “fire everyone.” It means redesigning roles around what’s now possible. It means identifying who in your business can operate at 3–5x with the right tools, and giving them those tools and the mandate to use them.
It also means being honest about who can’t or won’t make that leap. That’s the painful bit. But pretending otherwise is more painful in the long run.
2. Rip up your roadmap
I’d bet money that 50% of what’s on your product roadmap right now is now buildable in a fraction of the time, by a fraction of the team, for a fraction of the cost.
I’d bet another 20% doesn’t need humans at all. It can be agentified. Automated. Shipped overnight.
And I’d bet 10–20% of what’s on your roadmap shouldn’t be there at all anymore, because the underlying problem has either been solved by AI more elegantly than your planned feature, or it’s about to be.
So rip the roadmap up. Start again. Build the next 12 months around the world that actually exists now.
This is hard because roadmaps represent commitments. To customers. To investors. To your own team. Walking away from them feels like failure.
It isn’t. Sticking to a 2022 roadmap in a 2026 reality is the failure.
3. Reinvent how you hire
Your next ten hires shouldn’t be “more of the same.”
Stop hiring people who can do the job the way it was done in 2022. Start hiring people whose default operating system includes AI. People who think in agents. People who treat AI as a creative collaborator, not a calculator.
I’d rather hire one person who’s deeply augmented than three who aren’t. The output will be higher. The cost will be lower. The cultural ripple will be enormous.
This means rewriting job descriptions. Rewriting interview processes. Rewriting comp structures. Rewriting what “senior” even means.
It also means being honest with the people you already have. Some will rise to this. Some won’t. Both are fine. But pretending the bar hasn’t moved is a disservice to everyone.
4. Question every process
Every workflow. Every meeting. Every report. Every review cycle.
Ask one question: if a calm, capable agent could do this overnight, why is a human doing it on a Tuesday?
Most of the work happening inside most companies right now is work that no longer needs humans. Status updates. Meeting summaries. First-draft documents. Routine analysis. Customer support triage. Initial research. Pipeline reporting.
I’m not saying delete these things. I’m saying delete the human cost of doing them, and reinvest that human cost into the things only humans can do, judgment, taste, relationships, vision.
This is how you actually create the leverage that augmented companies have. Not by adding AI on top of human work, but by removing human work from the things AI can do, so humans can focus on what they’re uniquely good at.
5. Reinvest the savings - don’t pocket them
This is where most companies get it wrong.
When AI saves you money, the temptation is to take the saving. Lower the burn. Improve the margin. Tell investors you’re more efficient.
Don’t. Especially if you are scaling.
The companies that win the next decade are going to take every pound they save through AI and reinvest it into bigger bets. Bigger product ambitions. Bigger market expansion. Bigger talent. Bigger brand.
If your competitors are pocketing their AI savings and you’re reinvesting yours, you’ll be operating at twice their ambition with the same cost base. Within 18 months, you’ll be in a different league.
This is the leverage compounding I keep writing about. It only works if you have the discipline to reinvest, not extract.
6. Get radically honest about what only humans can do
Here’s the optimistic part.
The more AI takes over the routine, the more valuable the genuinely human work becomes. Strategy. Vision. Taste. Relationships. Hard conversations. Cultural leadership. Pattern recognition across messy, ambiguous situations. Trust.
These don’t go away. They become the highest-leverage activities in your business. Maybe the only ones that matter.
So in the rebuild, don’t just ask “what can AI do?” Ask “what is now uniquely valuable that humans can do, and how do I structure my entire company around protecting and amplifying that?”
The companies that get this right will feel different to work for. More space. More ambition. More agency. Less busywork. Less performative output. More real thinking, real decisions, real impact.
That’s the company I want to build. That’s the company I want to work in. That’s the company customers will actually want to buy from.
The hardest part isn’t the tech
Here’s what almost no one is talking about, and it’s the bit that keeps me up at night.
The biggest blocker to rebuilding for the world that exists now isn’t the tools.
It’s not the budget.
It’s not even the strategy.
It’s the skills gap inside your own team.
The numbers are damning. 71% of UK employees are using unapproved AI tools at work. Only 28% feel confident using AI properly. In businesses that have officially adopted AI, only 30% of staff actually use it.
That’s not adoption. That’s chaos.
You’ve got people quietly using AI badly. Outputs that look polished but no one can properly evaluate. Inconsistent standards across teams. Hidden compliance and reputational risk. And zero visibility into who’s actually capable versus who just looks confident in the meeting.
This is the silent crisis inside almost every pre-AI company right now.
The leadership team thinks the problem is “we need an AI strategy.”
The actual problem is: most of your people don’t have the skills to operate in the world that now exists, and you have no reliable way to assess or change that.
You can’t rebuild around a workforce that doesn’t know how to use the tools the rebuild depends on. You can’t think with AI as your operating system if your people can’t. And you can’t make the leverage gains you’re hoping for if half your team is quietly using ChatGPT to write emails and calling it transformation.
This is exactly why I co-founded AcademyAI.
Not as another course library. Not as another LinkedIn Learning clone. Not as a one-week bootcamp that produces certificates and zero behaviour change.
As an AI skills platform, built on cognitive science, embedded in the flow of work, with a live AI companion that coaches your team in real time as they actually work.
The premise is simple: most learning platforms sell content. We sell more capable people.
We do it through three things working together:
A learning platform that teaches AI fundamentals and role-specific AI skills (finance, sales, marketing, management, HR, compliance), built on actual cognitive science; spaced repetition, retrieval practice, desirable difficulty, so the skills actually stick.
A live AI companion that lives on your team’s desktop and coaches them in real time as they’re using AI in their actual work. Not a course they sit through once. A coach that’s always there, always improving them.
A community and verification layer so you can actually see who’s capable, who’s not, and who’s getting better, across every team, every function, every role.
It’s the missing infrastructure for operating in the world that now exists. The bit between “we bought ChatGPT licences” and “our people are genuinely operating at 3–5x.”
If you’re a CEO trying to rebuild your business for the world that now exists and you’re hitting the wall I just described, the silent skills gap, the polished outputs you can’t evaluate, the inconsistent capability across teams, that’s the gap we close. Quietly, at scale, across every function.
If that’s a problem you’re trying to solve, come and talk to us. We’re working with companies that have decided this is the moment to do something serious.
What this looks like inside a company
Let me make this concrete, because the abstract version is too easy to nod along to and ignore.
Imagine two companies. Same sector. Same size. Same revenue. Both founded in 2018.
Company A: still operating in the old world.
They’ve bought enterprise ChatGPT licences. They’ve run a couple of internal lunch-and-learns. They’ve added “AI-enhanced” to two product features. They’ve hired a Head of AI who’s mostly running a working group. The CEO talks about AI in every all-hands. The roadmap has an “AI initiatives” workstream. Nothing has fundamentally changed.
Cost base: roughly the same as 2022, maybe a bit higher because they’re paying for AI tools on top of everything else.
Output per person: roughly the same as 2022.
Roadmap velocity: roughly the same as 2022.
Team morale: declining slightly, because the best people can sense the inertia and are starting to look elsewhere.
Company B: rebuilt for the world that actually exists.
They spent six months in 2025 doing genuinely uncomfortable work. They restructured. The org chart is 30% smaller and significantly more senior. Every remaining role has been redesigned around AI augmentation. They’ve put a real skills programme in place, not a one-off training, but ongoing capability building with verification. They’ve ripped up the old roadmap and rebuilt it around what’s now buildable. They’ve reinvested the cost savings into bigger product bets and senior talent acquisition.
Cost base: 25% lower than 2022, despite higher AI tooling spend.
Output per person: 3–4x what it was in 2022.
Roadmap velocity: 5–10x what it was in 2022.
Team morale: high, because the best people can feel they’re working at the edge of what’s possible.
Now fast-forward 18 months.
Company A has held steady. Revenue’s roughly flat. They’re spinning their wheels, not declining, but not growing. The board is starting to ask uncomfortable questions.
Company B has shipped what would have been three years of roadmap in 12 months. They’ve launched into adjacent markets the old structure couldn’t have supported. They’ve poached the best talent from Company A. They’re growing 3–4x faster, with better margins, with a more energised team.
Now fast-forward another 12 months.
Company A is being acquired. Or quietly losing market share. Or going through painful restructuring. The board has replaced the CEO. The story being told to the new leadership is “we should have moved faster.”
Company B is the category leader.
This isn’t a hypothetical. This is the pattern I’m watching play out in real time, in real companies, right now. And the gap between Company A and Company B will be irreversible by the end of 2027.
Why most CEOs won’t do this
I want to be honest with you about something, because I think pretending otherwise is dishonest.
Most CEOs reading this won’t actually do the rebuild.
Not because they don’t see what I’m describing. Most do.
Not because they don’t have the strategic intelligence. Most do.
Not because they can’t afford it. Most can.
It’s because the rebuild is emotionally hard in a way that most senior leaders aren’t equipped to handle.
It means:
Telling people you’ve worked with for years that their role is changing fundamentally, or going away.
Admitting that some of the structures and processes you built and were proud of are no longer fit for purpose.
Walking away from roadmap commitments you made to customers and investors.
Letting go of the version of the business that made you successful, in order to build the version that will keep you successful.
Sitting with months of uncertainty while the rebuild happens.
Being okay with the fact that some senior people will leave because they can’t or won’t make the shift.
This is not what most CEO career paths prepare you for. Most senior leaders are pattern-matched to grow what exists, not to dismantle what exists in order to build something fundamentally different in its place.
The CEOs who will do this are the ones who care more about where the company will be in three years than how they look in the next board meeting.
That’s a small group. Always has been.
If you’re in that group, this is your moment. Genuinely. The window to do this from a position of strength; with revenue, with reputation, with a strong team to work with, is open right now. It won’t be open in 18 months.
If you’re not in that group, that’s fine too. But please be honest with yourself about it. Don’t pretend you’re rebuilding when you’re decorating. Your team can feel the difference. Your competitors can feel the difference. Your customers can feel the difference.
Running a company designed for a world that no longer exists is a death sentence. Slow, comfortable, and final.
A note on calm in the middle of all this
One last thing, because I write about it constantly and I think it matters more here than anywhere else.
If you decide to do this, to actually rebuild, the temptation will be to do it from a place of panic and stress.
Don’t.
Panic produces bad rebuilds. Reactive cuts. Performative restructures. Hires made out of fear. Strategy designed to look bold rather than be effective.
The CEOs who navigate this transition well are the ones who can hold two things at once: deep urgency, and complete calm.
Urgency about the timeline. Calm about the execution.
That requires a level of internal regulation that most leaders don’t have, and that no MBA programme teaches. It requires sleeping. Training. Thinking. Walking. Saying no to the noise. Protecting the space you need to make decisions of this magnitude well.
I’ve been writing about calm as a competitive advantage for a year now. It has never mattered more than it does right now. Because the decisions you’re going to make in the next 18 months will define your company for the next decade, and you cannot make them well if you’re operating from chaos.
So if you’re going to rebuild, build the calm to do it from too. They go together. They have to.
The line in the sand
If you take one thing from this:
The world your company was built for no longer exists.
You don’t need an AI strategy.
You need a company built for the world that does exist.
The CEOs who rebuild now; operating model, talent, products, costs, ambition, will be playing a completely different game by the end of 2027.
The ones who keep running their old company in a new world will be acquired, marginalised, or quietly gone.
I’d rather be in the first group. I bet you would too.
So do the harder thing. Now. While there’s still time to compound.
Either way: don’t wait for the next board meeting to start. The clock is already running.
Your old company isn’t coming back.
Build the new one.
Charlie - thank you so much for the positive feedback, thats really kind. Congrats on selling out of your company. Touch base soon.
Genuinely your best blog yet. And fascinatingly insightful having been on both sides of this shift / process in one of your investments. Niceeee AP. A must-read for any leader like, yesterday.