Complete AI Training
Start for Free

Your job's AI kit

Get your AI kit

Tell us who you are and what you do. We show you your kit right away and email you the link: skills, prompts, AI agents, MCP servers and courses for your job.

500+ jobs ready, and we make a kit for any other job. No payment needed to look.

Share

AI news ·

Big spend, small strategy: UK AI gains at risk from shadow tools and silos

UK firms are ramping AI spend to £15.9m this year, eyeing 32% ROI by 2027. Yet piecemeal rollouts and shadow AI are leaking value, data, and trust.

UK AI investment is rising fast - but fragmented adoption and shadow AI are cutting returns

22 October 2025 - London, UK. New research from SAP, produced with Oxford Economics, shows UK companies will spend an average of £15.94 million on AI this year, with investment expected to climb a further 40% over the next two years. The average firm reports a 17% return today (about £2.7 million) and expects ROI to reach 32% (£7.5 million) by 2027. Most executives (78%) expect a positive return within one to three years, and 52% say AI delivers value faster than any other technology.

Some of that return is forecast to come from agentic AI. UK businesses expect an 11% ROI from intelligent agents within two years, equivalent to £2.7 million, as agents begin to connect finance, supply chain, and workforce planning end to end.

The strategic gap: fragmented adoption

  • 42% say investment is piecemeal
  • 37% say it's department-led
  • 15% say it's ad hoc
  • Only 7% have a strategic, enterprise-wide plan

It's no surprise 70% of UK businesses are unsure whether AI is delivering its full potential. As one UK leader puts it, too many are still treating AI as a "technology project" instead of a business transformation. The opportunity is to rethink how the company operates, how people work, and how value is created for customers.

Shadow AI is a signal - but also a risk

Employees are moving fast. 68% of organisations say staff use unapproved AI tools at least occasionally. The costs are real: 44% have seen data or IP exposure and 43% report security vulnerabilities tied to shadow AI.

Training isn't keeping pace. 60% of companies say staff haven't completed comprehensive AI training, though 71% are starting to reskill or upskill. Leaders call for sanctioned tools, clear guidance, and safe spaces for experimentation. Psychological safety matters - employees need to ask questions, report issues, and learn without fear - while companies keep the right controls and ethical standards in place.

Value is showing up - if data, skills, and strategy click

  • 36% report significant improvement in decision-making
  • 34% see gains in customer engagement
  • 31% report faster time to value

These outcomes matter in a service-led economy where knowledge and relationships drive growth. But without a clear strategy, strong data foundations, and the right skills, enthusiasm won't translate into sustained productivity or competitiveness. As one executive noted, too many teams are "filling a shopping basket with tools" rather than linking data, applications, and intelligence into one system and building end-to-end use cases.

Executive takeaways: turn momentum into measurable impact

  • Set enterprise outcomes first. Pick 3-5 measurable targets (revenue, cost, cycle time, risk), assign owners, and fund against them - not isolated pilots.
  • Prioritise cross-functional agents. Start with high-leverage flows (order-to-cash, forecast-to-fulfil, hire-to-assign) that span finance, supply chain, and workforce planning.
  • Fix the data basics. Establish a common data model, quality checks, lineage, and access controls. Keep sensitive data out of prompts and enforce retention rules.
  • Replace shadow AI with approved choices. Publish an acceptable use policy, provide vetted tools, log usage, and red-team critical use cases.
  • Train by role, not by hype. Baseline everyone, then offer role-specific paths for sales, finance, operations, and IT. Consider structured programs to speed adoption. For curated options by role, see Complete AI Training.
  • Build a practical governance stack. Model registry, human-in-the-loop for material decisions, audit trails, and incident response. Helpful guardrails: NCSC guidance on secure AI development and the ICO's AI and data protection guidance.
  • Measure ROI and risk continuously. Track adoption, unit cost, cycle time, error rates, revenue lift, and model drift. Set clear "kill or scale" thresholds.
  • Tighten vendor terms. Require data handling controls, IP indemnity, uptime SLAs, security attestations, and breach notification commitments.

Read the full report

The Value of AI in the UK: Growth, people & data

Methodology

Oxford Economics fielded an online survey (July-August 2025) of 1,600 senior executives across eight global markets, including 200 from the UK. The sample covered midmarket firms (500-1,500 employees) and enterprises (1,500+), with respondents at director level or above.

[1] Respondents estimated in USD. GBP figures use a 0.7381 exchange rate (Bank of England, 1 September 2025).

Share