Industrial companies are still in the early stages of agentic AI and are conducting initial pilot projects. However, they should already be preparing for the organisation-wide rollout of this technology, otherwise they run the risk of getting stuck. IFS highlights the biggest pitfalls in the transition from the pilot phase to scaling.

Automated quality control, predictive maintenance, optimisation of material flow: AI agents that perform specific tasks autonomously without human input and make decisions independently promise industrial companies significant efficiency gains. Some companies have already launched pilot projects or even implemented their first AI agents.

However, the next step, which involves scaling up this technology, presents numerous challenges for industrial companies. They need to be prepared for these challenges, otherwise they run the risk of getting stuck in the transition phase from pilot projects to the widespread use of agentic AI. IFS, a leading provider of industrial AI software, explains the biggest obstacles:

Fear of change. When it comes to introducing technologies that drastically change their operations, industrial companies are often hesitant. This reluctance stems from concerns about losing control over important business processes and uncertainty about how these changes will affect workflows. These concerns are likely to be particularly pronounced when introducing autonomous systems such as AI agents.

Difficult to measure success. Without clearly defined success metrics, it is difficult for companies to evaluate the effectiveness of IT tools. With agentic AI, however, it is particularly challenging to determine how it specifically impacts key business outcomes such as productivity, costs or operational efficiency. AI agents often operate in complex, dynamic environments, which makes objective and comparable evaluation difficult.

Uncertainty about use cases. It is not immediately apparent which complex processes AI agents can effectively map. To find out, companies need in-depth expertise and a clear understanding of their internal processes. Otherwise, there is a risk that they will only use agentic AI for simple tasks and thus fail to exploit its full potential.

Lack of data readiness. Agentic AI relies on available, high-quality data. Preparing and providing data is a particularly big challenge for industrial companies, as they often have many legacy systems in use and their data landscape is highly fragmented. They need a data framework that can support a large number of AI agents that process huge amounts of data in near real time.

Resistance among the workforce. Since agentic AI completely automates certain tasks and takes over from humans, resistance from employees is inevitable. Although they can focus on their high-value tasks and support agents who need approval or are faced with uncertainty, they will undoubtedly have to give up a certain degree of autonomy in their work.

‘For a successful organisation-wide rollout of agentic AI, industrial companies need to make strategic investments and proceed methodically,’ explains Stefan Issing, Presales Director DACH at IFS. “They should define clear roles and business outcomes for the agents and get their data and infrastructure ready. To ensure a smooth transition, it is also important to invest in onboarding initiatives for employees and to form cross-functional teams at an early stage.”