AI is hollowing out the old office model.

London’s future lies in smaller, design-led spaces built for collaboration, not headcount.”

“This is not a person.” That disclaimer, used in a recent AI video demo, was meant to warn viewers about synthetic media. It could equally serve as a warning to office workers. Generative AI is no longer nibbling at the edges of knowledge work; it is swallowing whole tasks once thought untouchable. Reports, contracts, marketing campaigns, code – AI can already produce them.

For London, a city whose economy has long been powered by white-collar jobs, the implications are profound. The office, the skyscraper and the very fabric of the City are being reshaped. The disruption is not just about how many days a week we commute. It is about whether tomorrow’s billion-pound businesses will need offices at all.

Knowledge work under pressure

The old assumption was that automation threatened factory workers while professional roles were secure. Not any more. JPMorgan has warned of a “jobless recovery” in the next downturn as AI suppresses demand for clerical and graduate roles. Surveys show UK companies are more likely to invest in AI tools than in new staff. Klarna’s chief executive put it bluntly: “AI is capable of doing all our jobs, my own included.”

That prospect unnerves many. But for those who adapt, AI is already an amplifier. Employees who use generative tools report higher productivity and less stress. By 2030, it is estimated that 70 per cent of today’s workplace skills will have shifted. In other words, the office of the future will have fewer people, each doing more, supported by software rather than by juniors.

The nano unicorn thesis

Venture capitalists are preparing for an era of “nano” companies: one- or two-person firms that achieve unicorn status with the help of AI. OpenAI’s Sam Altman has predicted the first one-person billion-dollar start-up within a couple of years. This sounds like provocation until you look at the evidence.

Lovable, a Swedish AI coding platform, reached unicorn status in eight months with only 45 staff. Cursor, another developer tool, has hundreds of millions in revenue with fewer than 50 employees. Gumloop raised a Series A with two full-time staff, openly aiming to hit a billion-dollar valuation with a team you could fit round a dinner table.

Scaling in this world does not mean hiring. It means spinning up more cloud servers and licensing larger AI models. That is exhilarating for investors and founders, but destabilising for labour markets. If one person plus an AI stack can achieve what once required 200 staff, what happens to those 199 jobs?

Hybrid by default

London already reflects these pressures. Average office attendance remains more than a quarter below pre-pandemic norms. The working week has collapsed into three busy days and two quiet ones. Fridays are hollow, Mondays not much better. Companies are reducing space, subletting floors or exiting leases altogether. Canary Wharf’s vacancy rate has climbed to record highs.

At the same time, prime Grade A space in the City is still nearly full. What companies do keep, they want to be excellent: light, flexible, equipped for hybrid meetings and good coffee. The office is no longer a warehouse for employees. It is a tool for culture, collaboration and client trust.

The case for small, design led offices

If AI shrinks teams and pushes routine work into the cloud, the office must become more purposeful. That points towards small, design-led spaces rather than vast floorplates. These “nano offices” will act as clubhouses. They will be used two or three days a week, not five. Every square metre will be designed for impact.

The principles are straightforward but often ignored: right-sized footprints, acoustic comfort, natural light, greenery, varied settings for focus and collaboration, invisible technology, and sustainability by default. Not luxury for its own sake, but comfort and clarity that help people think faster and decide more quickly.

For landlords, the implication is fewer long leases and more modular, curated suites. Fit-outs must be adaptable. Services will matter as much as square footage: reliable technology, concierge-style support, hospitality touches. Neighbourhoods with mixed-use energy – Shoreditch, Fitzrovia, Clerkenwell – are already ahead. They offer what Canary Wharf struggles to supply: streets that stay alive after six o’clock.

London’s choice

London can cling to a model of towers filled with anonymous desks, or it can embrace this shift. Conversion of surplus space to housing, labs or cultural uses is already underway. But the greater challenge is cultural: accepting that the office is no longer compulsory. It must justify itself, not through presenteeism but through the quality of experience it offers.

Generative AI is forcing us to rediscover what the office was meant to be. Not a container for workers, but a space for ideas. Not a fixed cost, but a flexible asset. The winners in this transition will be the small, beautiful offices that people choose because they make work better.

For a city built on commerce, that is both a threat and an opportunity. If London adapts quickly, it can remain the global capital of knowledge work – even if the knowledge workers now bring fewer colleagues, more cloud servers, and a great deal of invisible software with them.