The increasingly polarised divide between AI advocacy and scepticism is interesting for a few reasons. While people are often dismissive or wary of new technologies, at least at first, the depth of feeling about AI is on a different order of magnitude.

Here's what I think could be behind this.

1. An entire generation has yet to experience genuine technology disruption

While it's true that we've seen more technological progress in the last twenty-five years than ever, it's also fair to say that the last two decades have been relatively orderly and pedestrian compared with the often spiking, lurching tech advancements we saw in prior decades.

For example, take the smartphone, the pre-eminent technological artefact of our time, which arrived with huge fanfare and then promptly reverted to a pattern of incremental improvement, with a steady drumbeat of consumer-friendly incrementalism that enables market analysts to accurately predict the next couple of years' worth of new smartphone features well in advance. And it's been much the same for the rest of our modern tech inventory. Broadband connectivity. Camera resolution. Cloud apps. Social media platforms. We've become accustomed to each segment delivering a small handful of new features or marginal improvements every year or so.

If the underlying architecture of today's tech world was largely defined by the mid-2000s, what followed has been more about evolution than revolution. Therefore, anyone who came of age with technology after roughly 2000 has never experienced a paradigm shift; they've only experienced acceleration within a stable one.

In this context, the threat model for AI is not "Uber for [insert category]" or "Desktop software but in a browser," but something much harder to comprehend: intelligence itself becoming a commodity. And, frankly, that does not sit easily with an evaluative framework that two decades of gradual, tamed progress have etched onto our brains. It's therefore not surprising that this is disorienting and discomforting (and contempt-inducing) for many.

Interestingly, Gen Z (those born between 1997 and 2012), at least according to recent reporting, are both the biggest users of Gen-AI and its greatest sceptics.

2. The product logic is inverted

For more than forty years, the best product thinking has held to a single organising principle: start with the customer problem and work back to the technology. Jobs, Bezos, and every serious practitioner in the tech world faithfully espoused this law: understand the customer's need first, then build the technology to serve it.

But the Generative-AI era has flipped this script at scale. The capability arrives first — vast, general and only partially understood — and the search for problems comes after. Entire product categories are being constructed not from observed human need but from the desire to deploy a capability that's demo-ready. When products feel like solutions hunting for problems, people sense it.

3. The industry had already spent its trust

AI did not arrive on a blank slate. It arrived after two decades of surveillance capitalism, algorithmic addiction, platform monopoly, and recommendation systems optimised for engagement and marketing at the cost of everything else. The resulting grievances are legitimate, and so when an industry with that track record announces that its next product is the most transformative technology in a generation, some degree of hostility is understandable and frankly, warranted.

4. "Software brain" has made it personal

Underneath the distrust sits something more specific. Writing in The Verge last week, Nilay Patel beautifully coined a term for it: "Software brain". In his framing, it describes a particular habit of mind that fits everything into algorithms, databases, and automated loops — one in which, as Patel puts it, Zillow is a database of houses, Uber is a database of cars and riders, and every human preference is a data point to be captured and acted upon.

Patel's argument is that software brain has always run the tech industry, but AI has turbocharged it, giving more businesses, in more sectors, the ability to automate more of their operations than at any point in history. The result is a fundamental disconnect between how people with a software brain see the world and how regular people live their lives.

As a long-term software brain adherent, to me, that diagnosis rings uneasily true, but it understates what makes this particular moment so challenging for some. AI does not just reflect software brain, it deploys it directly onto lived experience at scale, and without consent. The chatbot that will not transfer you to a human. The hiring screen that filters your CV before anyone reads it, or the recommendation engine that always feels slightly off. These are not neutral tools. They are software brain applied to everyday life, and the cumulative effect is a loss of agency, the feeling that the system is optimising for something, and that something is probably not you.

Thus far, the tech industry's response has been wide of the mark. Sam Altman thinks it's a marketing problem, and so last month bought a daily YouTube AI news channel. Satya Nadella says the industry needs to earn social permission. Both treat the hostility as a perception gap that better messaging could close.

But, as Patel says, you cannot market people out of their own experience. The animosity will not recede until a new perspective forms, and people's evaluative instincts start working again — and until the industry reckons honestly with what software, at scale, actually does to people.

In the meantime, the hostility is worth reading as a signal. Not that AI is overhyped, although that is most certainly the case in some quarters, but that something with the texture of real, disruptive innovation has arrived.

And after twenty years of relative comfort, this new sense of discomfort may just be what real disruption feels like.

I didn't grow up in poverty, but it was always within shouting distance. In 1968, I was the firstborn to young, working-class parents, born into a one-bedroom Glasgow council tenement, where for my first three years I had a bed tucked into a corner of the kitchen. My world was an extended family comprised entirely of manual workers; what we lacked in formal education or social capital, we compensated for with love, encouragement and grit.

For a kid with my statistical life chances, the technology industry lowered a social mobility ladder that I grabbed with both hands. Like many of my generation, ignited by an all-consuming passion for computing, I taught myself to code, to build, to navigate the new digital world that unfurled before me in the 1980s. I would go on to spend more than thirty years putting that fluency and passion to good effect, achieving far more in my career than that tenement kid could have dreamt of.

Therefore, I believe from lived experience, in what technology can do for people. But the hope for technology that was so abundant in the 1980s has given way to uncertainty. And as the parent of a young adult daughter, the concern I hear from other parents about their children's future career prospects in an AI era doesn't feel abstract to me at all.

The ladder

Technology has been one of the most powerful engines of social mobility in history. The democratisation of access to tools, information, and markets has done more for people on the outside than almost any policy intervention of the last fifty years. Working-class kids with access to a low-cost computer at school or in the home could teach themselves to code. A small business owner without capital could reach customers nationally, not just locally. A first-generation student without connections could learn anything.

AI looks set to extend this further, equipping individuals with capabilities that previously required teams, budgets, and credentials. A junior analyst with access to AI has the kind of research and synthesis firepower that would have been unimaginable a decade ago. A sole trader can now produce marketing, legal summaries, financial modelling, and customer communications that once required expensive specialists. That is genuinely exciting, and it would be dishonest to pretend otherwise.

But the drawbridge is going up

But a countervailing force is equally real. The people best positioned to benefit from AI are those who already have proximity to it — the educated, the connected, the credentialled. AI literacy, like financial literacy before it, is not evenly distributed. And access to the best tools, the most capable models, and the networks that know how to deploy them is concentrating rather than dispersing.

There is also a structural problem. The knowledge worker roles that technology created over the last thirty years, the ones that served as the new mobility ladder for millions of people without inherited wealth or formal education, are precisely the roles most exposed to the potential of AI displacement. Copywriters, analysts, junior coders, paralegals, researchers. The ladder that technology lowered is at risk of being pulled up by the same hand. Most damaging prospect of all: we may be about to automate entry-level jobs, the very apprenticeship phase that allowed people like me to climb.

Power and value are concentrated in a small number of companies and people, and the governance structures seem thin. And the people making the decisions about AI's direction are, almost without exception, already at the top of the system they are reshaping.

When I speak to other parents, the overwhelming theme is one of profound uncertainty about their children's futures and career prospects. Meanwhile, AI technology leaders casually throw around near-term forecasts of massive job reductions and disruption, with no one grasping the nettle to close the gap between AI's pace of progress and the prospect of a generation of kids about to be left behind. This is the precise opposite of the "BBC Micro" moment: the 1980s UK government initiative that put affordable computers into schools and helped give a generation, including me, its first foothold in the digital economy.

Some will say that AI is the great leveller; that a smartphone gives a kid in a Glasgow scheme the same firepower as a CEO in a London boardroom, but I think that might be a convenient myth. There is a world of difference between being a user of a technology and being a master of it. If we only teach the next generation how to swipe a screen or ask a chatbot a question, we aren't giving them a way out; we're just turning them into high-tech consumers in an economy they have no power to shape. Access to a tool is not the same thing as the agency to change your life with it. Without a real, national effort to teach the under-the-hood fluency I was lucky enough to find in 1984, we are just giving our kids a front-row seat to their own displacement.

What needs to happen

This is not an argument against AI, it's an argument about choices, and right now, the people making those choices are not doing so with sufficient urgency or intent.

Addressing this requires two parallel efforts: building foundational AI literacy into schools for the generation just starting out, and creating genuine pathways for adults already in the workforce to reskill, retrain, and if necessary, reinvent their careers entirely. Both matter, but they are different problems requiring different solutions.

The second track is harder and less comfortable to confront. We are familiar with the idea that there is no longer such a thing as a job for life. If AI's trajectory continues as many predict, that phrase could start to look quite optimistic. What we may actually be facing is no such thing as a career for life. If that's the case, front-loading education into the first two decades of a person's life and expecting it to sustain the next four is a model built for a world that may not exist for much longer. Lifelong learning and adult re-education need to become a fixture of working life, not a safety net for the unlucky.

Governments set the conditions in which either outcome becomes more likely. To be fair, the UK government has recognised the urgency. Its AI Skills Boost programme, launched in January, aims to provide free AI training to every adult in the country, with an ambition to upskill ten million workers by 2030. That is welcome, and it is more than most governments have done. But upskilling today's workforce and building a ladder for tomorrow's generation are two different things. A programme aimed at helping existing workers use AI for drafting text and managing administrative tasks is not the same as ensuring the next generation of working-class kids has the foundational literacy to shape the AI economy, not just survive it.

AI literacy needs to enter the school curriculum as foundational, not optional and taught from primary school upwards, as the equivalent of reading and numeracy for the generation now in class. Broadband and device access are still not universal. The gains from AI productivity need to be returned to the people most exposed to its disruption.

A model for this exists, and we know why it worked. In the late 1970s, the UK was confronting an uncomfortable reality: in business and industry, the silicon chip was transforming the way people worked, threatening jobs and leaving the country dangerously uncompetitive - sound familiar? The government recognised that without deliberate intervention, an entire generation would be left without the skills to participate in — let alone shape — the new economy. In response, in 1981, the UK government worked with the Cambridge-based Acorn Computers and the UK national broadcaster, the BBC, to put over a million BBC Micros into schools across the country, along with extensive TV-based educational programming — a deliberate, state-backed infrastructure for a new kind of opportunity. It was the ladder made policy. What is needed now is the equivalent: a new BBC Micro moment; a national AI literacy initiative, built for the pace and scale of what is coming, and aimed squarely at the people who need it most rather than those already closest to it.

What can we do?

Technology did not give me a ladder by accident. It gave me one because a government, a broadcaster, and a computer company decided that a working-class kid in Glasgow deserved access to the future. That was a question of intent then.

And with the pace things are moving, it is most definitely a question of intent now.

We've observed decades of tech adoption begin inside large organisations and then slowly trickle down to SMBs and individuals, but I'm beginning to suspect the reverse is happening with AI.

For every tale of woe I hear about inconclusive ROI and fruitless AI pilots within large organisations, I hear about a freelancer or self-employed person — unencumbered by corporate IT and procurement policies — who has lashed together various AI tools, workflows, and cloud services to utterly transform their personal productivity. It seems the secret has been hiding in plain sight all this time: productivity is personal, not organisational.

Previous enterprise shifts — ERP, CRM, Cloud — were built on the process as the unit of value. They required top-down, command-and-control rollouts to prevent left-hand/right-hand chaos and waste. But AI's unit of value is the individual: their specific thinking, creativity, drafting, and judgment. A technology that augments the idiosyncratic way a human mind works will always be adopted more readily by people acting in an individual capacity.

The data bears this out. Among large companies with more than 250 employees, AI adoption has dipped from a peak of 14% to 12%¹. Among freelancers, it stands at 75%².

Why does scale make it worse?

When it comes to rolling out AI, large organisations responded with the only playbook available to them: standardise, govern, procure, roll out, measure. Copilot licences were purchased, lunch-and-learns were scheduled, and 42% of companies abandoned most AI initiatives in 2025³, up from 17% the year before.

The problem contradicts everything large organisations believe about their own advantages: diseconomies of scale. In most enterprise tech deployments, scale amplifies, but with AI, the reverse seems true.

The coordination overhead of deploying a personalised technology stack to hundreds or thousands of diverse workers compounds with every layer of governance. When an organisation forces a single-vendor configuration on its staff — as corporate procurement policies usually dictate — it sacrifices the very thing that makes AI valuable today: the high-fidelity fit between the tool and the individual user's unique cognitive style.

If economies of scale built the enterprise, diseconomies of scale are frustrating its ability to harness AI.

Headcount reduction is not a consequence of AI adoption. It looks more like a precondition.

The current wave of large-scale tech layoffs reshaping global organisations like Oracle, Amazon, and Block is not, in the main, caused by AI replacing workers, as the media usually reports it. Instead, it is being driven by boards who have concluded that organisation-wide AI adoption cannot be coaxed, encouraged or even mandated.

Instead, it will be forced by circumstance.

The re-org, redundancy programme, or RIF (Reduction In Force) is the lever usually yanked when large organisations need to enact a major change in direction or strategy. CEOs can demand that their managers prudently constrain headcount, rationalise and manage costs and encourage the adoption of AI in their teams, but it's more efficient to simply remove the optionality.

So, my sense is these cuts are coming ahead of the AI productivity gains, not because of them. When there are fewer people to absorb the work, necessity will become the mother of invention, and those who remain will be forced to find new ways to do it — with AI tools.

The self-employed are building personal AI operating systems while large organisations are still negotiating procurement terms. How that gap closes is the more interesting question.

Sources: Fortune ¹, 2727 Coworking ², S&P Global Market Intelligence ³.

Twenty-five years ago, I was a tech blogger. Before LinkedIn, before Twitter, before Substack. Over the course of three or so years, I wrote a good number of posts where I would think-out-loud about what the Web, the then-nascent Web 2.0, early social media and what web services might do to the world of software, business and society. The early 2000s were an exhilarating time of intense learning, building, fun and unbridled creativity, with debates and ideas about where the Web was taking us literally zinging off the walls.

Fast forward to 2026, and these days I spend a decent chunk of my time meeting with early-stage tech founders, venture capitalists, CEOs of mature tech businesses, and a few private equity partners, all of whom inform and shape my perspective. I've also personally invested in a number of SaaS and AI-native start-ups over the last five years, and so I have some first-hand skin in the game, too.

In short, while we grapple with what AI is, arguably more pertinently what it isn't and where, whatever it is, is taking us, I have an uncannily familiar feeling in my gut again.

A feeling, just like twenty-five years ago, that we're back in a full-blown technology revolution, and suddenly everything is exciting again. And even if you're a fervent AI Kool-Aid refuser or sceptic, I'm confident that if you were to soberly weigh all the AI arguments and viewpoints, for and against, you would at least concede that some kind of corner has been turned.

So, this is probably the closest to the kind of thinking-out-loud I used to do twenty-five years ago. These are not predictions; some may be more like propositions, and they are definitely not fully baked. They're just reflections, hunches or observations, and as such, they're fully open to challenge and critique. After all, what's the point of an opinion if you never change it?

1. The strongest signal right now

The single most important signal in Q1 2026 isn't a benchmark, a valuation, an analyst research paper or a polemic. It's behavioural. The engineers who least need AI assistance are the ones most enthusiastically embracing it.

Here is perhaps the most telling data point in the entire AI debate. Some of the world's most capable software engineers, the people who could write elegant, complex, production-grade code in their sleep, are now among the loudest advocates for AI-assisted development. And many of them will tell you, without embarrassment, that they no longer write code in the traditional sense.

This is not a story about AI replacing mediocre developers. This is about elite practitioners voluntarily and enthusiastically changing how they work because the output is better and delivered faster. When the best people in a field restructure their own practice around a new tool, that is not hype. That is a leading indicator.

It's also important to note that this particular advantage is not limited to only pioneering developers and start-ups operating at the edge of tomorrow. I've spoken with CEOs of mature, market-leading software businesses who are blending (and increasingly mandating) AI-assisted coding into their organisations.

And professionals in other domains, clinging onto the belief that AI is unlikely to automate chunks of their daily workflows any time soon, don't understand how insanely complex software engineering is. Or, rather, was.

2. Good enough to be dangerous

The AI debate is fixated on the wrong threshold. AGI shouldn't be the bar for mainstream utility. Instead, SGI or Sufficient General Intelligence, LLM-based AI that while narrow, is capable enough, in enough places, enough of the time, may present a significant opportunity for pragmatic transformation and operational leverage, and possibly sooner than most expect.

The most common dismissal of large language models follows a predictable line of argument. Critics acknowledge the impressiveness, then pivot: these systems don't truly understand. They're not rational, nor sentient. They lack grounded world models. They confuse correlation for causation. Plus, they hallucinate. And therefore, the argument goes, today's AI is nowhere near ready to tackle white-collar work.

The first part of that argument is probably correct. True AGI (whatever that is) will require something LLMs fundamentally don't have: a causal, deterministic and rational model of how the world actually works. Not probabilistic statistical patterns across text, but relational frameworks of consequence, time, and physical reality. This will require new reasoning approaches or architectures we haven't built yet. The people making this point are not wrong.

But they're drawing the wrong conclusion from it.

There is an inconvenient truth here that we may not have fully reckoned with: we would likely be surprised, and possibly a little unsettled, by just how little intelligence is required to complete a good-ish number of discrete white-collar tasks. A surprising proportion of professional work was never bundled into human roles because it required Einstein-level intelligence eight hours a day. It was bundled there because humans were the only general-purpose processors available. Our work was never as continuously complex as our salaries might have implied.

And so even if LLMs only ever attain Sufficient General Intelligence, possibly as little as 10% to 20% (though I suspect for some roles, a lot more), of a human's capability across a given role or domain, this may still represent significant operational gain. If conventional software were the hand tool, it amplified what a skilled human could do, but the human still had to show up, apply judgment, click the buttons, and do the work. Then SGI may be closer to the early machinery that emerged at the beginning of the Industrial Revolution.

SGI won't require world models or sophisticated reasoning. It requires clearing the bar in enough places that the old assumptions stop holding. That bar hasn't been cleared yet, but at the speed things are progressing in early 2026, it may not take much. Agentic AI, systems that can plan, act, and iterate across multi-step tasks, could be the unlock. Not a new architecture. Not a breakthrough in super-intelligence. Just current capabilities, given enough autonomy and freedom to run.

And after all, our memories are short. We've happily accepted imperfect software for more than 50 years, with all its foibles, bugs and functional gaps, as it evolved and gained utility over time, so perhaps it's about time we cut LLMs some slack, too.

None of this means the end of white-collar work. The industrial revolution didn't eliminate human labour; it redistributed it, reduced the headcount needed for certain tasks (which, of course, will weigh on society), but it also created new categories of work that hadn't existed before. SGI is likely to follow a similar pattern. Fewer people will be needed to process, triage, click, draft, and summarise. But the humans who remain in those roles will operate at a different level, spending less time on routine tasks and more on judgment calls that LLMs can't handle, and therefore still require genuinely human qualities: context, relationships, ethics, and accountability. The workforce will shrink in some places and shift in others. That is disruptive enough to take seriously without reaching for apocalyptic conclusions.

The critics may be right that AGI is a long way off. They may be wrong about how much that actually matters.

3. The Death of Distance, revisited

AI has all but eliminated the distance between having an idea and building it. That is genuinely transformative. But it has not reduced the distance between a good idea and a bad one — and in the age of vibe coding, the difference between the two gets built and shipped faster than ever before.

In 1997, economist Frances Cairncross published The Death of Distance, arguing that the internet would make geography economically irrelevant. She was right; we didn't end up with a different Google for every town and city on the planet, as conventional wisdom might have led us to presume at the time, but with a global baseline of instant information. In 2026, we are witnessing a second, more profound collapse: the death of the distance between intent and execution.

Vibe coding and AI-native development have effectively reduced the "build time" of an idea to near zero. A principle that was captured perfectly this week by Azeem Azhar, author of the excellent newsletter, Exponential View:

"The projects I build with my AI agents weren't impossible before — they were sitting in a queue for months because the cost of explaining them to anyone exceeded the cost of just not doing them. That cost has now collapsed." — Azeem Azhar

For the first time in history, you do not need to know how to build something in order to ship it. A founder can prototype without an engineer; a marketer can deploy a functional tool without touching a line of code. But there is a trap hidden in this efficiency. While AI has eliminated the distance between having an idea and building it, it has done nothing to reduce the distance between a good idea and a bad one. In fact, it has made the bad ones more dangerous.

This brings to mind a warning from the dawn of the personal computing era. Adam Osborne—the brash, British-American pioneer who created the first commercially successful portable computer, the Osborne 1, in 1981—once amusingly noted:

"People think computers will keep them from making mistakes. They're wrong. With computers, you make mistakes faster." — Adam Osborne

Osborne was speaking in the early 1980s, a time when the mere act of computerising an accounting ledger felt magical. His point was that the machine didn't grant wisdom; it only granted velocity. Today, with AI, that velocity is no longer just "faster"; it is exponentially faster. We won't be just making mistakes at the speed of a microprocessor; we'll be making them at the speed of thought, and, worryingly, potentially in a greatly enlarged theatre of operations, where the impact of one simple design mistake could have massive ramifications for thousands or even millions of businesses and users.

When execution is difficult or costly, bad ideas usually die in the friction of development. When execution is instant, the world is suddenly flooded with high-fidelity, functional noise. We may be entering the era of "Confident Mediocrity", where the output looks professional, but the underlying logic is hollow or flawed.

Analysts Ben Thompson and Benedict Evans recently discussed this tension on the always-insightful Stratechery podcast. To paraphrase them, do we really want David Brent—the blissfully incompetent manager from the BBC mockumentary, The Office—vibe-coding their own payroll or ERP system?

The risk is not that the AI will fail to build what Brent asks for. The risk is that it will build exactly what he asks for, even if it's nonsense. If you give a high-powered execution engine to someone with poor judgment, you don't get a better product; you just get "wrong" at a higher velocity. AI is a force multiplier, but zero multiplied by a thousand is still zero, so it quickly becomes a farce multiplier instead.

Anish Acharya, general partner at VC firm Andreessen Horowitz, framed the opportunity differently on a recent 20VC interview with Harry Stebbings:

"If you look at SaaS spend today, if you look at IT spend overall, it's 8-12% of enterprise spend. So, even if you vibe-coded your ERP and your Payroll, with all the kind of risks and dangers that that entails, you're going to save 8-12%? You have this innovation bazooka with these models, why would you point that at rebuilding payroll, or ERP or CRM? You're going to take it and use it to extend your core advantage as a business, or you're going to optimise the other 90% you're not spending on software today." — Anish Acharya

As the technical barrier to entry evaporates, the signal in any industry is shifting upstream. In the old world, the bottleneck was the ability to build — Technical Fluency. In the new world, the bottleneck is the ability to direct intent — Intellectual Taste. The most successful "vibe coders" and AI-assisted leaders won't be the most technically gifted; they will be the most expertly informed and the best at articulating and communicating intent.

They are the ones who can look at a generated output and see the subtle "hallucination of logic" or nonsense that a novice would miss. In the age of instant execution, the only remaining competitive advantage is judgment. We no longer need people who can 'do the work' as much as we need people who know what work is actually worth doing.

4. The Ghost UI

For twenty-odd years, I've used my own personal heuristic to gauge whether a B2B software business was any good or not. We used to judge software by its elegance. We may come to judge it by its absence.

My philosophy is very simple. If a software company is proud of its product, and if it's any good, then it prominently displays it as a hero image on its website's front page, and glamorous product screencaps are peppered throughout the rest of the content. Conversely, less product-led software companies that adorn their homepages with stock photography of executives looking pensively into the distance, or who offer "request a demo" buttons, are usually concealing vast amounts of UI complexity, an ugly or badly outdated design, or a product that needs so much costly professional services effort to function at all, that there is in fact nothing meaningful to actually show.

After three decades, I am not sure my heuristic survives in an agentic world, where a UI is not a feature — it is redundant. Every minute a user spends navigating a menu is a minute the software has failed to do its job autonomously. The best AI-native products will be the ones you interact with least — they'll manifest as a notification, an auto-drafted reply, a completed task that was waiting for you when you arrived in the morning. The heuristic is inverting: a carefully designed interface is no longer a sign of sophistication. It is becoming an anachronism.

Which, of course, throws into the bin the entire discipline of choosing and selecting software, along with 5-star software reviews for good measure. Agents with the appropriate levels of autonomy will select, in milliseconds, which skills and connections to employ in their Ghost UIs to get stuff done — and that will be a far more logical and clinical affair than a beauty parade.

5. Per-seat pricing and price list innovation

The per-seat licence was built for a world where humans did the work. AI is breaking that assumption. But before writing off enterprise software incumbents, take note of another of my heuristics: When innovation stops in the product, it moves to the price list.

The last 12 months of SaaS valuation repricing has been driven by a number of factors; 1) analysts and investors are no longer certain about the near to mid-term future, and markets hate uncertainty, 2) nor do they deeply appreciate the complexity, depth or range of variability in customer value between one given SaaS product over another, so they bundle everyone into 'at risk', and 3) larger enterprise SaaS businesses who price per seat e.g. ERP, CRM, HR, etc. start to look shaky if AI begins to displace employee headcount.

The per-seat licence is an elegant model for a world where humans do the work. As a company grows, it hires more people, and enterprise SaaS revenue grows in lockstep. It is simple, predictable, and has made a generation of enterprise software businesses very wealthy.

But it was designed for a time when the unit of productivity was a person. AI is breaking that assumption. If an autonomous agent does the work of ten analysts, the customer's productivity skyrockets, and the incumbent's per-seat revenue collapses by 90%. Legacy SaaS vendors are now financially incentivised to keep users busy rather than effective. Every AI copilot they launch is carefully calibrated to keep a human in the chair, because a truly autonomous product makes the per-seat contract an absurdity. AI-native entrants will have no such constraint because they'll typically price on outcomes delivered or token usage, not headcount.

This is a real structural problem. But it would be a mistake to assume that incumbents will simply sit still while their model and revenues erode.

Faced with existential pressure on the per-seat model, enterprise SaaS vendors will do what they have always done: repackage, rebundle, and reprice with considerable speed, creativity and innovation. Consumption-based models, outcome tiers, agent licences, platform fees, and AI add-on SKUs are already appearing. Some of this is genuine model evolution. Some of it is the per-seat model wearing a different coat. And the investor community, already struggling with product literacy, will find pricing innovation at least as hard to read as product innovation.

The price list is the last line of defence. And in enterprise software, it is a surprisingly well-fortified one.

6. In SaaS, the context is the moat

While AI might easily replace 'thin' SaaS apps that merely shuffle data around, it cannot replicate the context, the deep organisational memory and complex workflows embedded in 'thick' systems of record.

Double-clicking on the variable nature of value in SaaS, and therefore the AI-risk profile of one product over another, it turns out that context is everything.

On one hand, you have 'thin' SaaS apps that are more like UI-wrappers sitting on top of a SQL database. Data goes in, gets shuffled around, produces reports, or is possibly sent off via an API for processing elsewhere. These are more susceptible to disruption from AI.

Then you have 'thick' systems of record. A good example would be your accounting app, where not only is transactional data ingested, recorded and reported, but where your business processes and the discrete workflows of your business (at least relating to accounting in this example) are codified and defined behind every button and discipline. These systems of action provide a critical element of context; they understand the why behind an organisation — its rules, its politics, its memory — and are the only thing standing between that organisation and the chaos of having to relearn itself from scratch.

You can probably work out which side of the risk register each category of SaaS app belongs on. If you want a deeper take on where systems of record go in an AI-infused world, Jerome Gouvernel's article this week is an excellent place to start.

7. AI-forcing and multi-region SMB

European SMB software is highly fragmented by local regulations, culture and practices, making pan-European expansion cost-prohibitive compared to the more standardized enterprise market. This has historically prevented native vendors from achieving global scale. However, AI-assisted coding may finally reduce the costs of localisation and compliance, allowing these companies to overcome the structural barriers that have long limited Europe's tech growth.

From a B2B software perspective, Europe is a nightmare, particularly so in SMB. It may be unified by currency, free passage, and multiple laws and harmonised regulations, but at the ground level of business, it's incredibly fragmented. This is one reason why there are few (if any) SMB software vendors that have successfully gone pan-European.

Enterprise software is different. The needs of large companies and multinationals have much more in common with other multinationals than with their own countryfolk, and so they are more readily disposed to adopting enterprise products like Salesforce, Oracle, NetSuite, Workday, etc.

Not so with SMBs. First, they're often less operationally mature or formalised and therefore more easily squeezed into less configurable, predefined products that level them up operationally; they often reflect the culture and regional preferences of their local business communities, unlike their larger multinational siblings; and they rarely need features that cater for multi-country trade. And finally, they vastly outnumber large businesses.

Therefore, if you're an SMB focused B2B SaaS vendor in France, Germany or the UK, not only is there no demand from your ICPs for multi-regional features, but the sheer cost and complexity of building, localising product, country-specific integrations, banking, reporting and compliance, never mind standing up go-to-market teams in multiple countries, is just eye-wateringly cost-prohibitive. Plus, the likely domestic TAM outside your front door is measured in the millions, so why bother?

There are a number of structural outcomes that result from this. Few European native B2B software vendors ever reach global scale, at least organically (it's feasible via M&A), and most multi-regional SaaS vendors are not European. This has profound implications for Europe in the next twenty years, many which were articulated perfectly in a white paper published by Boardwave in 2024.

It's possible, if not probable, that AI-assisted coding leverage may finally tame the prevailing cost-inefficiency equation of taking SMB tech multi-region.

8. The rise of the all-terrain leader

Most leaders are shaped by where they came from — sales, finance, product, engineering. That origin story used to define their operational ceiling. AI is about to change that.

In retrospect, one ability that I'm certain enabled me to operate more effectively than my peers during Xero's early years was the range of competencies I utilised across multiple operational and functional disciplines.

Most CEOs typically accumulate experience and develop skills within a specific domain — whether that be sales, finance, technical, or operations and that background tends to then shape or flavour their overall ability and effectiveness as leaders.

Fortunately for me (and Xero), I was a somewhat atypical odd-ball in being able to operate with sufficient competence and confidence across a wider range of disciplines than most leaders I've encountered: think general management with a capital "G". Sales, people, investors, marketing, product, strategy, communications, finance, hiring, culture — never at expert depth, but usually deep enough to solve problems without needing to call in aircover from elsewhere in the business. That mattered enormously in the early Xero years, when anyone I might have called on for help or support was likely asleep in New Zealand while I was awake in the UK.

So if something needed doing, I usually just did it myself. I believe this autonomous way of working, the resulting execution speed and the breadth of understanding I could wield across the whole business were a genuine competitive advantage at an important growth stage.

AI is about to make range available to anyone. The constraint was never intelligence or ambition — it was access to the right knowledge, at the right moment, explained in the right way. Naval Ravikant, in a podcast last month on AI and the future of work — beautifully titled A Motorcycle for the Mind and clearly riffing on Steve Jobs in 1980 when he spoke about personal computers being bicycles for the mind — described AI as the most patient tutor imaginable — and one with relevant context that can meet any learner at exactly their level and explain and apply anything, in as many ways as it takes, until they genuinely understand it. Not dumbed down. Not overwhelming. Precisely calibrated to the edge of what you already know. That kind of learning was previously available only to the privileged few with access to exceptional teachers or expensive advisors. Now it's available to any leader willing to ask.

This will be the engine behind the all-terrain operator. A founder who came up through engineering can now think fluently about or finely orchestrate their go-to-market processes. A sales-led CEO can interrogate and shape their product roadmap. The cognitive overhead of operating outside your native domain — the gap between knowing enough to act and needing to hire someone who does — is vanishing. Breadth was once a happy accident of how a career unfolded. It will soon become learnable.

To conclude, as I sit here in early 2026, the numbness in my legs is fading, but the feeling in my gut is stronger.

If the first era of the Web was about access (finding information) and the second was about connection (social and SaaS), this third era is about synthesis. We are moving from a world where we had to learn the language of machines to a world where machines have finally learned ours.

But as the distance between having an idea and building it disappears, we are left with a startlingly clear view of our own judgment. When the "how" becomes trivial, the "why" becomes everything. The real winners of this revolution won't be those with the most powerful AI, but those with the best "Intellectual Taste" — the ones who know what work is actually worth doing in the first place.

Twenty-five years ago, everything felt new, and the walls were bouncing with ideas. It's taken a quarter of a century, but that feeling is finally back.

There's still a great deal of debate and skepticism about where LLMs are taking us. This is natural and not entirely dissimilar to when the web went mainstream thirty years ago, and the same with cloud apps around twenty years ago.

It is also worth reflecting on how long it took those shifts to progress from niche nerd topics to broader awareness and then adoption. Technological inertia is powerful; it takes about a decade after the intellectual arguments are won before we see majority adoption. Even if it were perfect, fully baked, and 100% ready to go today, it would still take several years for AI to be comprehensively adopted and digested by more than a minority of businesses.

So I think it would be a mistake, therefore, to disparage the current state of language model-based AI, as I occasionally hear, on the basis that it doesn't deliver sustainable economic returns to those who try to deploy it.

I personally suspect we're about 1% of the way into wherever LLMs are going, and it's far too early to size it, just as Thomas Watson, the founder of IBM, famously did when he said he thought there was "…a world market for maybe five computers".

However, my time in software spans three decades plus, which means I'm pathologically incapable of moderating my impulse to map prior shifts to the present one and attempting to proverbially peer around the corner.

While generalised agentic AI business applications are still percolating, and AI investors huff and puff themselves into a bubble — what if it's not a bubble? — much of what I'm reading and hearing points to AI-assisted coding having turned a corner in the second half of 2025. Despite AI agents making the headlines for mostly the wrong reasons, a quiet revolution is happening in the IDE where the sentiment among developers is shifting rapidly:

"I've never felt this much behind as a programmer... I have a sense that I could be 10X more powerful if I just properly string together what has become available over the last ~year and a failure to claim the boost feels decidedly like skill issue." — Andrej Karpathy

"The quality of work produced by Claude... regularly blows my mind. I cannot and do not want to go back to that infuriating, head-scratching, hair-pulling, sloth-like programmer productivity of, er, 2 years ago." — Olly Heady

"I believe the AI-refusers regrettably have a lot invested in the status quo...They all tell themselves that the AI has yet to prove that it's better than they are at performing X, Y, or Z, and therefore, it's not ready yet. But from where I'm sitting, they're the ones who aren't ready." — Steve Yegge

The implications are profound. Collapsing production costs and simultaneously leveraging productivity in software development will spark a number of significant changes, not least a likely explosion in the number of new products coming to market.

You're gonna need a bigger metaphor

The Cambrian Explosion was occasionally employed as a somewhat grand metaphor by VCs to describe what happened to software sometime around 2010, when the practice of building desktop software was finally obsoleted by the general availability of reliable, low-cost cloud services. Building software for the cloud fundamentally altered and flattened the prevailing cost of production, distribution, and operations in SaaS model businesses, quickly spawning thousands of new SaaS startups.

As great as the shift to the cloud was, it's possible, if not probable, that AI-assisted coding will bring about a much bigger shift.

Anish Acharya, a General Partner at Andreessen Horowitz, makes a compelling case in his article "The Future of the Web is the History of YouTube," arguing that the structure of the software industry could be about to undergo a huge metamorphosis, akin to the impact YouTube has had on video content creation and distribution.

In barely 20 years, YouTube has succeeded in all but eliminating the cost of producing and distributing video content (and monetising it), resulting in a colossal number of mainstream and niche-interest channels and content producers, with a staggering 700,000 hours of new content being uploaded every day.

"We may see hyper-personalized applications on the web, for much smaller audiences. This is tremendously liberating: software no longer needs to be practical... It just needs to have a good idea behind it, and a couple of people who understand its value." — Anish Acharya

Working backwards, YouTube's impact on the media industry was more or less predicted in 2008 in Clay Shirky's book Here Comes Everybody - The Power of Organizing Without Organizations.

Mass amateurisation, the democratisation of expertise or unadulterated chaos?

Which brings us back to pattern recognition. Just as skepticism was abundant during the early days of the web and the cloud, overly fixating on the risks of AI-generated code misses the forest for the trees. The collapse of production costs is not a variable; it is a trend line.

Yes, we will be awash in niche, "good enough" software, many of which will go nowhere or fail. Yes, relying on unproven, micro-scale vendors carries risk. But to focus solely on the potential for chaos is to make the same mistake Thomas Watson made with the computer. We are witnessing the democratisation of engineering, where the barrier to entry drops from "millions of dollars" to "a good idea."

As I said, I believe we may well be only 1% of the way into all this. If the YouTube era taught us anything, it's that while mass amateurisation brings chaos and noise, it also unleashes creativity on a scale previously impossible. The next ten years won't just be about who can write code, but who can navigate a world where coding is no longer the bottleneck—and who can filter the signal from the noise.