Time wasted on repetitive tasks

Industrial companies are wasting 41% of their employees' time on manual and repetitive tasks, according to a study by The Futurum Group that surveyed 664 decision-makers from six industrial sectors in North America and Europe. Nearly two days per week are spent on operations that require no human judgment, and 77% of the surveyed decision-makers have delayed or abandoned initiatives due to lack of capacity.

Operational capacity deficit

This gap between the work that needs to be done and the available resources is called the operational capacity deficit. It becomes increasingly costly to ignore, especially as experienced engineers and operators are retiring and not being replaced quickly enough. The arithmetic of hiring has reached its limits.

Focus on freeing existing capacity

The study suggests that the focus should be on freeing up existing capacity rather than funding new capacity. Digital workers, defined as agents that monitor operations, make decisions, and execute tasks end to end—only requesting human intervention when judgment is required—are presented as a solution.

Current use of automation

Two-thirds of the surveyed companies plan to allocate a budget for this in the next twelve months. However, actual use remains very low: 10% currently operate an AI functioning mainly autonomously, and only 5.7% grant it full autonomy to act. This gap does not reflect technological skepticism but a requirement for predictability that few platforms currently satisfy.

Balanced automation in practice

The binary choice between full autonomy and refusing automation does not match industrial reality. Deployments that are in production distribute the work: agents handle routine tasks, and employees retain control over decisions that require discernment. With an agent platform, a company could automate 60% of agent transactions end to end, with the remaining 40% involving human control and approval steps.

Key requirements for automation

Three requirements recur among industrialists who have successfully implemented automation: native integration into existing company systems, governance and traceability built from the start, and an escalation model that only escalates genuine exceptions.

Real-world example of integration

A Nordic electronics manufacturer, committed to deploying procurement agents across its thirteen sites by the end of 2026, abandoned automated process automation because each new supplier introduced its own rules, leading to unmanageable maintenance costs. Its agents are now integrated into existing software, leaving current safeguards and approval processes unchanged.

Automation without data cleanup

The dominant reflex is to launch a data cleanup project before any automation. However, documented deployments in the study suggest the opposite approach. None of the six surveyed companies waited for a sanitized data environment. At the electronics manufacturer, an agent identified a ten-year-old product reference error that had gone unnoticed by manual checks.

Financial impact of automation

The implementation pace changes the nature of the financial decision. A European supplier of industrial solutions generalized its supplier order management agent across the company in ten weeks. An American food and beverage industrialist recovers 27 hours per week previously spent on manual processing of supplier orders. At these time scales, the investment compares to a budget cycle, not a multi-year transformation plan.

Specific AI agents over general-purpose AI

A general-purpose AI will not fill this deficit. It is digital workers designed for specific tasks, integrated into existing systems, that return capacity to teams rather than teaching them to cope with shortages. What separates a pilot from a production deployment is a number—the share of transactions that still go back to a human—and its ability to decrease month after month.