57,000 jobs. 142,000 tech layoffs. AI's labour impact just showed up in the official numbers.

What changed in the first half of 2026

The scale of AI capability deployed into business workflows in H1 2026 was unlike any previous six-month period. Claude Sonnet 5 became the default Anthropic model. OpenAI's GPT-5.6 reached enterprise customers. Microsoft's Copilot Wave 3 integrated AI agents into M365. Google's Gemini reached general availability as a business tool. AI coding tools achieved 97% developer adoption at large companies.

The jobs data is the first official confirmation that this deployment wave has reached the labour market at scale. It is not a sudden event — it is the accumulation of many businesses making the same decision: replace administrative headcount with AI capacity, reduce entry-level hiring, and redirect budget to AI infrastructure. When enough businesses make the same decision simultaneously, it shows up in aggregate payroll data.

The UK's specific exposure

A fifth of all tasks performed by UK workers are exposed to AI automation — a higher proportion than in any other advanced economy. The reason is the UK's industry mix: finance, law, media, professional services, and consulting are all heavily represented, and all rely heavily on knowledge work — the category most directly displaced by current AI tools.

UK unemployment stood at 5.2% in early 2026 — a post-pandemic high — while wage growth has slowed to approximately 3.8%. Approximately one in five UK firms has already reported staffing reductions attributable to AI. Around one in ten firms has moved beyond generic AI tools into deeper, bespoke AI integration — and those firms are where the productivity and cost advantages are concentrating.

What this means for service businesses — not a threat, a signal

Most UK service businesses — chimney sweepers, tradespeople, stove installers, consultants, independent retailers, local professional services — are not in the industries where AI is replacing large numbers of employees. Their exposure is not "AI will take our jobs" — it is "AI is reducing the cost of administrative work, quote generation, customer communication, and follow-up for businesses that use it, creating a growing productivity gap between AI-capable operators and manual operators."

In a service business context, AI does not replace your skilled labour. A chimney sweep, a plumber, a sparky — those jobs require physical presence and skilled hands. But the administrative work around those businesses — quoting, booking confirmation, follow-up reminders, review requests, customer communication — can be substantially AI-assisted. The businesses that have built those systems are reducing their admin cost by one to three hours per day. The ones that have not are still paying for that time.

The workflow AI insight buried in the jobs data

One signal from the broader labour market analysis is particularly relevant for service businesses: the market is now rewarding workflow AI — tools that finish bounded tasks with human review (support routing, quote generation, compliance drafting, research prep) — over general-purpose chatbot use. This is the AIFA message exactly. The AI use cases with the clearest ROI in 2026 are not "ask the AI anything" — they are "automate this specific process, review the output, send."

If you have tried AI tools and found them useful but not transformational, the reason is almost always that you tested them as chatbots rather than as workflow tools. Workflow AI — where you define the specific task, template the input, review the output, and act on it — delivers consistent value. Chatbot AI delivers occasional value when you happen to ask the right question.