5 Minutes
Imagine opening your email and finding the first draft of that contract, the project plan and the budget notes already polished — not by a junior analyst, but by software. That scenario stopped sounding like science fiction in a recent Financial Times interview with Mustafa Suleyman, who now leads Microsoft's artificial intelligence unit.
Suleyman’s claim is blunt: many of the routine, desk-based tasks performed by knowledge workers — whether you’re a lawyer, accountant, project manager or marketer — will be automated within the next 12 to 18 months. He didn’t hedge with caveats about decades of gradual change. He said performance for a broad set of professional activities is approaching human levels, and that the effect will be swift.
Within a year and a half, many administrative tasks could be routinely handled by AI, not humans. Short sentence. Big implication.
The timing and tone of that prediction have stirred fresh anxiety. Investors reacted sharply after Anthropic unveiled a new workplace-focused model called Claude Cowork; markets tumbled as traders imagined legal and advisory workflows being replaced by cheaper, faster software. The worry is two-fold: people losing jobs, and entire business models — companies that sell specialised admin tools or bill for routine expertise — being squeezed.
Microsoft’s own executives have already been suggesting the change is underway. Satya Nadella has said more than a quarter of some of the company’s code is now produced with AI assistance, and a Spotify co‑CEO recently claimed their platform’s coding needs are largely handled by AI. The headline here isn’t mystery: developers are adopting tools that write and refactor code, generate tests, and surface bugs. The nature of engineering work is shifting from typing lines to supervising, debugging and designing architectures.
That shift is visible in the day-to-day. Many engineers report using AI to scaffold new features or to auto-complete repetitive code paths. They spend less time on boilerplate and more on systems thinking. But this is not a simple replacement of human judgment with a black box. Instead, it’s a change in the relationship between creator and tool — one that has dramatically evolved in just a few months.
Still, important questions remain. Are the outputs reliable? Do they speed teams up or slow them down? Early studies and field reports are mixed. Some organisations see productivity gains. Others find developers end up reviewing AI-generated code multiple times, which can negate time savings and introduce new kinds of errors. In administrative tasks, automated drafts often require heavy human editing before they can be used in critical contexts like legal filings.
And then there’s the economic picture. If large swathes of routine work become automated, where does value accrue? Will incumbents who control AI platforms capture the upside, or will customers demand lower prices for services that used to be labour‑intensive? Software vendors that make money from subscription tools for niche workflows could face margin pressure if a general-purpose AI can replicate their core features.
None of this implies an immediate moral panic. Complex problem solving, negotiation, strategy and certain forms of creativity still depend on human context and relationships. But the workplace of today will not look the same a year from now. Roles will be redesigned. New jobs will appear. Some will vanish. The real test will be whether organisations can harness these tools to free people for higher-value work — and whether regulators, educators and leaders move fast enough to manage the social consequences.
So, what should professionals do? Learn to collaborate with these systems. Audit their outputs. Treat AI as a powerful assistant that needs supervision rather than an infallible replacement. That’s less dramatic than a takeover, but far more realistic — and immediately practical.
Will that be comforting to someone worrying about their next performance review? Maybe not. But it is where the hard conversation starts: not about when the machines will win, but how humans will adapt, supervise and profit from the tools now reshaping office life.
Leave a Comment