Before and After AI
A Timeline of Everything That Changed and a Toolkit for What Comes Next
Last week, Jack Dorsey cut Block’s workforce nearly in half. Not because the business was failing. Block had just posted a gross profit of $2.87 billion for Q4, up 24% year on year. Dorsey did it because he believes a smaller team, armed with AI tools, can do more and do it better. Over 4,000 people walked out the door. The share price jumped 18%. Dorsey predicted that within a year, most companies will make similar structural changes.
That event tells you everything about where we are. Around 245,000 tech jobs were cut globally in 2025, with roughly 70% originating from US-headquartered companies. AI was directly cited in nearly 55,000 of those. By early March 2026, over 50,000 more had gone, running at 800 per day. Amazon cut 16,000 in January. Microsoft shed 9,000. Meta trimmed 1,500. These are not struggling companies. These are some of the most profitable businesses on the planet, restructuring around a technology that lets them operate with dramatically fewer people.
Before AI, When Companies Hired Their Way to Growth
The line falls in late November 2022, when OpenAI released ChatGPT. Everything before that date now feels like a different era. Between 2019 and 2022, major technology firms nearly doubled their headcount, fuelled by pandemic-era digital acceleration and cheap capital. Alphabet grew from 119,000 to over 186,000 employees. Meta went from 45,000 to 87,000. Block nearly tripled. More people, more output, more growth.
Knowledge work was labour-intensive by design. A junior lawyer reviewed contracts line by line. A marketing team of fifteen produced the content an AI can now generate in an afternoon. Customer service at Klarna was handled by 3,000 outsourced agents across phone, email and chat. Entry-level roles were plentiful, sometimes tedious, but they served as the training ground through which the next generation learned their craft. The implicit deal was clear: do the unglamorous work, absorb tacit knowledge about how the business operates, and you will progress.
The digital economy ran on a stable model. Google dominated search. Advertising revenue flowed through a well-understood ecosystem of programmatic buying, keyword bidding and display networks. Content marketing was a growth industry. SEO agencies thrived. The attention economy, the business of capturing human eyeballs and selling them to advertisers, was the foundational logic of the entire internet. The whole model assumed a human would type a query, scan a results page, click through to a website, and see an advertisement along the way. Enormous industries were built on the permanence of that assumption.
After AI, When Fewer People Became the Strategy
ChatGPT hit 100 million users faster than any application in history. The first wave of changes was additive: companies bolted AI features onto existing products, experimented with copilots, and talked about productivity gains. But the subtractive process began almost immediately. By early 2024, Klarna’s AI assistant was handling two-thirds of all customer service chats, the equivalent work of 700 full-time agents, resolving enquiries in two minutes instead of eleven, across 23 markets and 35 languages. Salesforce slashed its customer support workforce by around 4,000 after its own AI agents proved capable of handling the work. Y Combinator startups reported MVP development time collapsing by 60%. Entry-level hiring at the fifteen largest tech firms fell 25%. The experiment phase was over.
Klarna’s story also illustrates the limits. Customer satisfaction dropped. CEO Siemiatkowski admitted the company had prioritised cost over quality. They rehired human agents and shifted to a hybrid model. But the direction of travel remained. In 2025, Intel shed 15,000 people. Verizon cut 13,000. Dell disclosed 13,000 in its SEC filing. Gartner now predicts that 20% of organisations will use AI to flatten their structures through 2026, eliminating more than half of current middle management positions. A Harvard study tracking 62 million workers found junior positions shrinking at companies integrating AI. Stanford research showed workers aged 22 to 25 in AI-exposed fields experiencing a 13% relative decline in employment, even as older colleagues saw gains. The Dallas Federal Reserve confirmed the mechanism: AI replicates codified knowledge but not tacit knowledge, substituting for juniors while complementing seniors. The European Central Bank’s earlier research suggested displacement during the deep learning boom was overstated. The generative AI wave is proving to be a very different animal.
The pre-AI assumption that more humans meant more capability has been inverted. The post-AI assumption is that fewer humans, augmented by AI, can deliver equal or greater output. Whether this proves universally true is almost beside the point. The companies that matter are acting as though it is. And that is enough to change everything.
The Toolkit, Five Ways to Stay Relevant in the Post-AI Workplace
First, build a defendable skill stack. No single qualification protects you. A law degree will not save you when AI reviews contracts faster than a junior associate. The value lies in the specific intersection of your skills, experience and domain knowledge. Think of it as a Venn diagram. Individually, each element is replicable. But the combination, the unusual overlap of industry knowledge, client relationships, technical fluency and cross-disciplinary communication, is much harder to automate. Research into how computational thinking is being commoditised confirms that the premium is shifting from writing code to understanding what needs to be built and why. The most valuable person in the room is increasingly not the one who can write the code, but the one who knows what the code should do and can deal with a difficult client who keeps changing their mind.
Second, stay relentlessly current. PwC’s 2025 Global AI Jobs Barometer found that workers with AI skills command wage premiums up to 56% higher than their peers. Only 16% of workers have what Forrester describes as high AI readiness. You need working fluency with these tools, not a PhD. Know what Claude, ChatGPT and Gemini can actually do when pointed at real work in your field. This applies whether you are a software engineer or a tradesperson. The pace of progress in physical AI means nothing is permanently safe, and the wisest course is to upskill now while you have the luxury of doing it proactively.
Third, keep real-world skills alive. AI disrupts the digital layer, but the underlying human needs persist. If AI agents bypass search engines, the channels through which you reach human decision-makers may need to shift entirely. Television, events and direct mail may get a second life as channels that bypass the AI intermediary and reach humans directly. The platforms change, the channels change, but the fundamental act of reaching another person and convincing them of something remains the same. The skills that feel most old-fashioned, reading a room, building rapport, closing a deal over dinner, may turn out to be the most future-proof.
Fourth, become a trend spotter. The old career advice was to develop deep expertise in a stable field and let compound experience do the work. That assumed the field would remain stable long enough for the compounding to pay off. In an environment where entire sectors are being reshaped by AI within eighteen months, that assumption no longer holds. Skate to where the puck is going. If you are in ad-tech, the puck is moving away from search. If you are in customer support, it has already arrived. Block’s employees found out on a Thursday. By Friday, investors were celebrating. The market will not wait for you.
Fifth, invest in the human interface. Execution is getting radically cheaper. Understanding, the messy work of figuring out what actually needs to happen, retains its value. The jobs that survive AI will be the ones that involve navigating ambiguity, managing relationships, and making judgment calls where the data is incomplete and the stakes are personal. If your work primarily involves producing outputs that can be specified in advance, AI is coming for it. If your work involves the messy, deeply human process of working out what the right output would even be, you are stronger than you think.
The IMF estimates that roughly 40% of global employment will be affected by AI, with advanced economies facing the greatest disruption. The pre-AI world is not coming back. Build your stack. Stay sharp. Watch the horizon. And invest in the things that make you irreplaceably, distinctively human. The machines are getting very good at everything else.