Industry: 66% of companies plan to invest in digital agents

While 66% of surveyed decision-makers plan to invest in digital agents in the next twelve months, only 10% of companies currently use artificial intelligence that operates primarily autonomously. Trust, integration with business systems, and maintaining human control remain key challenges.

Manufacturers must accomplish an increasing volume of tasks with teams whose resources remain limited. In factories, supply chains and service activities, the retirement of experienced employees exacerbates this strain, while a significant portion of working time remains devoted to manual and repetitive operations.

A study conducted by Futurum Research for IFS, surveying 664 decision-makers in industrial companies across Europe and North America, estimates that nearly two-thirds of respondents believe 41% of their employees' time is consumed by these tasks. This situation fuels what the authors describe as an "operational capacity gap": the difference between the work companies want to accomplish and the human resources actually available.

Software capable of executing complete processes

Unlike an artificial intelligence assistant tasked with producing a recommendation or responding to a request, a digital agent is designed to execute a series of actions within a business process. It can monitor an operation, apply rules, make a decision within a predefined framework, update management tools, and escalate exceptional situations to a colleague.

These software agents can be used in order processing, purchasing, inventory management, maintenance planning, and delivery tracking. Their goal is not necessarily to replace human teams, but to free up their time for tasks requiring technical expertise, negotiation, or judgment.

The study nevertheless reveals a significant gap between investment intentions and the current maturity of applications. While two out of three companies plan to invest quickly, confidence in fully autonomous AI remains very low.

Human supervision remains essential

For the companies surveyed, scaling up depends on three main conditions: direct integration with existing business software, traceability of decisions and actions, and a human supervision system for exceptions.

The preferred model involves entrusting repetitive, clearly defined tasks performed in large volumes to digital agents. Employees retain the authority to approve sensitive or unusual situations. This structure allows the company to maintain the approval rules, responsibilities, and safeguards already in place.

Trust, therefore, appears less as a question of pure artificial intelligence performance than as a governance issue. Industry players want to be able to understand the decisions made, identify the data used, and interrupt or correct a process when necessary.

In construction, a resource planning challenge

Construction companies are among the six sectors covered by the study, alongside manufacturing, energy and utilities, aerospace and defense, transport and logistics, and telecommunications.

In the construction industry, the intended uses primarily concern resource and schedule planning. Material delays, manual coordination of stakeholders, and numerous adjustments can cause a chain reaction of delays in operations. Digital agents could monitor supplies, flag potential delays, or suggest schedule adjustments, while leaving the final decision to site managers.

This approach could also help reduce the time spent on administrative operations, in a sector facing recruitment challenges and the need to preserve available technical expertise.

IFS advocates a model combining automation and human control

With its IFS Loops agent platform, the vendor claims to be able to fully automate 60% of the transactions handled by its digital agents. The remaining 40% would involve a human review or approval step. However, these figures represent a general overview of the solution by IFS and may vary depending on processes, internal rules, and the quality of available data.

The agents are integrated into the software already used by companies. Approval processes and management rules can therefore be maintained, while cases falling outside the defined framework are referred to the appropriate staff.

Initial deployments in purchasing and supply

Several IFS clients are already using these tools in production. The food group KLN Family Brands reports having saved 27 hours per week on supplier order processing. These results are based on feedback gathered from client companies and do not constitute an independent evaluation of the solutions' performance.

Electronic component manufacturer Kitron Group is deploying digital agents dedicated to the process from purchase request to order across its 13 sites. The company anticipates full deployment in less than seven months, without modifying its existing rules or approval processes.

During this implementation, a digital agent also reportedly identified a product reference error dating back approximately ten years. This case illustrates the potential ability of these tools to detect certain inconsistencies during process execution. However, it does not allow us to conclude that a digital agent systematically exempts the company from prior work on the quality and governance of its data.

"The operational capacity deficit represents a major challenge for industrial operations. When a purchase order, a maintenance schedule or a supplier delivery is delayed because the teams simply do not have enough hours in the day, the consequences result in production stoppages, missed deliveries or unused equipment.

To bridge this gap, general-purpose AI isn't enough. What's needed are digital workers designed for a specific task and integrated directly into the systems that already power the business. This allows teams to reclaim capacity, rather than simply coping with a lack of resources. Our clients are demonstrating the relevance of this model in production, not just in pilot projects.

Somya Kapoor, CEO of IFS Loops

Large-scale adoption still needs to be confirmed

The results show that interest in digital agents is growing rapidly, but their widespread adoption remains contingent on several factors: data reliability, security, integration with existing systems, measurement of return on investment and acceptance by teams.

The study was commissioned by IFS and supplemented by in-depth interviews with six client companies already using its solutions. The figures therefore primarily reflect the perceptions of the decision-makers interviewed and the feedback from the vendor's customers. They do not constitute an independent measure of productivity gains achieved across the industry as a whole.

For companies, the challenge will now be to identify processes that are sufficiently repetitive and structured to be automated, while clearly defining the decisions that must remain under human responsibility.