AI & Work · 2031–2036
How Will AI Change Small Teams by 2035?
AI can save time, but useful gains depend on the workflow. Explore today’s adoption data, the economics of automation and three scenarios for the next decade.
3 min read · Published 2026-09-20 · Watch the full episode · 15 min ↓
The question
How will AI change the way small teams work over the next five to ten years? The most useful starting point is the work inside a job: searching records, preparing proposals, answering routine requests and deciding what needs human attention.
FXNumbers expects the biggest changes to come from connecting those tasks into better workflows. A faster first draft matters less if the rest of the process still depends on manual checks and missing information.
Three numbers to start with
20% — EU enterprises with at least ten people employed used AI in 2025, according to Eurostat. Adoption ranged from 17% among small enterprises to 55% among large ones. Using an AI tool does not mean that a business has automated its operations.
1 in 4 — The ILO estimates that a quarter of global employment is in occupations with some exposure to generative AI. This measures tasks that could change, not the number of jobs expected to disappear.
39% — Employers surveyed for the World Economic Forum's Future of Jobs Report 2025 expect this share of workers' existing skills to change or become outdated by 2030. It is an employer projection across several economic trends, not an AI-only outcome.
When does automation pay?
Consider an illustrative team that spends 100 hours a week on a recurring workflow. AI saves 30 hours of drafting and searching but adds 10 hours of checking and handling exceptions. The net gain is 20 hours.
At €30 per hour, that represents €600 of weekly capacity before software, integration and training costs. It becomes useful when the team can serve more customers, clear a backlog, improve quality or reduce overtime. Capacity is not automatically a cash saving.
The business case depends on time, quality, repeated demand and operating costs. Measuring accepted results is more informative than counting how quickly a tool produces an answer.
Three paths toward 2031–2036
Gradual adoption: assistants improve individual tasks, but fragmented data and manual handoffs limit the overall gains.
The FXNumbers base case: digitally mature businesses connect more of their workflows. Software prepares routine work; people handle exceptions, relationships and accountable decisions. Small teams can expand their reach without growing every function at the same rate.
Accelerated change: reliable agents complete longer assignments with less intervention, lowering the cost of launching products and delivering services. This depends on dependable execution in real businesses, not model demonstrations alone.
These are conditional scenarios, not measured outcomes or assigned probabilities. Competition and demand will influence whether gains become lower prices, higher wages, profits or shorter working hours.
What to watch
Five indicators can help distinguish progress from enthusiasm: business adoption, output per paid hour, cost per accepted result, hiring patterns and access to training.
If output improves while review costs fall, the accelerated path becomes more plausible. If usage grows but useful results do not, the gradual path deserves more weight.
What this means for a small team
Start with one repeatable workflow. Record its time, quality and cost, introduce assistance, then compare the accepted results. For individuals, the valuable combination is domain knowledge, AI fluency and judgement: knowing which problem to solve and recognising a strong answer.
The next decade may bring a different division of work between people and software. The pace will depend as much on how organisations redesign their processes as on what the technology can do.
Sources
Eurostat — Use of artificial intelligence in enterprises, 2025 data.
ILO — Generative AI and jobs: A 2025 update.
World Economic Forum — The Future of Jobs Report 2025.
Watch the full episode
Explore the examples and reasoning behind this article.