Businesses must support workers as they adopt AI agents over time
The article argues that businesses need to support employees’ ongoing adoption of agentic AI, rather than treating one-off training as enough. It says the distinction matters because technology may work as intended while failing to deliver business value if it does not change how people work.
Gartner expects agentic AI use to rise from 17% of businesses now to about 60% within two years. WalkMe’s 2026 report, based on 3,750 respondents, says 40% of digital transformation spending underperforms expectations, and employees lose an average of 7.9 hours a week to workplace friction. The article also highlights a gap between leaders’ estimates and actual tool use: executives estimate 35 applications where the average is 661, and 21 AI tools where the count is 80; 45% of workers said they had used unapproved AI recently, with 36% of those using confidential company data.
- Agentic AI adoption is expected to grow sharply.
- Poor adoption can leave technology spending below expectations.
- Unapproved AI use creates visibility and data governance challenges.
New here? Start with this
Agentic AI systems are software tools capable of carrying out work tasks with minimal human supervision. Organisations across industries are increasingly deploying these systems, with adoption rates expected to grow substantially over the coming years.
Simply introducing new technology to staff through training sessions is frequently insufficient for businesses to realise its value. A system may work perfectly whilst failing to generate genuine business benefits if it does not fundamentally change how employees approach their work.
Research reveals a significant disconnect between how business leaders perceive their workforce's tool use and the actual reality. Employees often adopt systems independently—sometimes without approval—whilst a large proportion of technology spending fails to deliver the expected improvements to business performance.
Both sides, in good faith
The strongest fair case each way — we don't pick a winner.
The case for
Organisations must provide ongoing support because technology alone cannot deliver business value without corresponding changes to how people work. The research is telling: 40% of digital transformation spending underperforms, employees waste nearly 8 hours weekly on workplace friction, and executives dramatically underestimate tool proliferation (35 estimated versus 661 actual), creating governance blind spots. This knowledge deficit permits risky unsupervised AI use with confidential data. Continuous support through change management, mentoring, and iterative learning closes capability gaps, ensures security compliance, and realises the promised ROI that one-off training cannot achieve.
The case against
Rather than requiring continuous organisational support, companies should establish robust governance frameworks that empower workers to drive their own adoption. The 45% of workers already using unapproved AI demonstrates genuine appetite to innovate and adapt independently. Extensive hand-holding creates costly dependency and organisational inertia. Instead, businesses should invest in clear policies, security controls, and performance incentives that guide responsible adoption whilst preserving worker autonomy. This approach distributes responsibility appropriately, recognises employee agency, and may encourage more sustainable, self-directed learning than imposed support programmes.
AI Celebrity Entertainment Technology
Read the full article at the source →
Originally published by The Register as “Embedding the human factor into AI agent adoption”.