Most small and mid-sized businesses (SMEs) have yet to leverage AI as a day-to-day asset, despite accumulating evidence that generative AI could transform a significant portion of knowledge work. Research by McKinsey suggests that current generative AI technologies have the technical capacity to automate tasks consuming between 60% and 70% of employees’ time, particularly in decision-making and administrative functions. However, many organisations have yet to harness this capability effectively in their day-to-day operations.
Key Takeaways
- AI can automate 60-70% of employee tasks, yet most SMEs haven’t tapped this potential
- Governance without action creates “shadow AI” where employees use tools without oversight
- Start small with mundane tasks rather than attempting large transformation projects
- The EU AI Act creates both challenges and opportunities through sandboxes and frameworks
- Success requires viewing AI as a capacity multiplier, not just a cost centre
2025-2026: Governance Over Outcomes
Attention among industry leaders has shifted towards AI policies and compliance rather than significant improvements in speed, cost, or quality.
Toolkits for SMEs provided by organisations like the OECD and national entities, such as Canada’s Federal AI Toolkit, prioritise governance and risk management. However, this has often resulted in organisations developing extensive restrictions rather than fostering an environment of innovation and enablement.
Surveys from 2025 revealed a growing trend towards AI policy implementation, yet very few SMEs had integrated AI on a large scale across core processes, leaving much of its productivity potential unused.
This emphasis on control led to “shadow AI” use, where employees experiment with tools like ChatGPT without formal company policies, resulting in fragmented use, ineffectively managed data flows, and uneven productivity gains. According to recent surveys, nearly a quarter of companies replaced some roles with AI in recent years, without robust governance frameworks.
Organisation, Not Tech
By 2026, the primary challenge remains organisational design rather than technological capability.
Many SMEs view AI as an IT and compliance issue, managed by teams more focused on risk mitigation than operational improvement. This risk-first approach is crucial but incomplete-it tends to produce rules rather than transformational workflows.
The economic advantages are clear: Brookings research suggests that generative AI has the potential to disrupt more than 30% of workers by altering at least half of their tasks, illustrating that much cognitive work is ready for redesign rather than mere optimisation.
What Adopters Did Differently
Some SMEs used governance as a foundation rather than an end goal.
Applying brief, risk-aware playbooks aligned with frameworks like the EU AI Act, they strategically targeted productivity impediments such as manual reporting and standardised customer queries. When employees were properly trained, research indicated significant performance gains, with AI-augmented workers completing tasks up to 10% faster than expert benchmarks.
These businesses did not aim for substantial transformation projects but started by automating mundane tasks. As tangible improvements in time savings and quality were realised, AI adoption and skills grew organically, reflecting broader findings that exposure to generative AI enhances both task performance and adaptability.
The 2026 Widening Gap
This performance gap is anticipated to widen through 2026 and beyond.
Removing even a single role’s worth of repetitive work via AI automation can rapidly justify AI investments. At the same time, regulatory pressures are intensifying, with the EU’s AI Act entering into force and SMEs at risk of compliance shocks and missed opportunities.
However, regulation is also creating support structures rather than just hindrances. The AI Act introduces sandboxes allowing SMEs to test systems under supervision. Those employing purpose-built governance tools have reported lower compliance costs, supporting the view that well-crafted oversight is compatible with rapid experimentation.
How to Move from Paralysis to Business Advantage
The main hindrance now is perception, not technology. Many leaders see AI as a mere expense without considering its potential to eliminate routine tasks. Organisations must view AI as a capacity multiplier, focusing on tangible improvements.
Key strategies include:
- Redesigning major operational bottlenecks rather than dispersing pilots.
- Establishing clear, efficient rules to manage data risks without stifling experimentation.
- Training teams for visible, measurable usage rather than concealed shadow tools.
- Engaging leadership to integrate AI into core operations.
Ready to Overcome AI Paralysis?
Don’t let your organisation fall behind. The gap between AI adopters and non-adopters is widening rapidly. EXPRE can help you:
- Assess your current AI readiness
- Identify quick-win automation opportunities
- Develop practical, risk-aware AI strategies
- Train your teams for effective AI adoption
As regulations tighten, the emphasis will be on well-documented yet progressive utilisation. The 2025-2026 experience emphasised the pitfalls of prioritising control over outcomes; success in 2026 and beyond will come to organisations striking a balance by fostering responsible experimentation and measuring success through faster, cheaper, or improved operations.
