Application of virtual AI agents in production process parameters setting and real-time production planningTABLE 1 Use case description1 Production process parameters setting Achieving optimal setting of machine parameters (such as temperature and pressure) is a key goal for manufacturers across all industries. The complexity of optimizing various process parameters under specific external influencing factors forces manufacturers to rely heavily on operator experience. AI agents are transforming this process to improve overall equipment effectiveness.2 Real-time production planning Real-time production planning is critical for manufacturers to meet demand, reduce lead times and optimize resource use. However, the complexity of balancing capacity, inventory, labour and external factors often requires manual adjustments by experienced planners. AI agents are transforming this process by streamlining decision-making and improving flexibility and responsiveness to changing conditions – ultimately enhancing overall production efficiency. Knowledge agent Parameter knowledge agents harness machine and process parameters and production output data (such as quality validation, material properties, and maintenance and quality reports) to identify optimal equipment settings for enhanced machine performance. The agent can be activated by workers via voice input or can raise alerts when predicting deviations. It continuously refines its analysis based on worker feedback. It also assists decision-making by estimating the potential value and cost impact of potential adjustments.Planning knowledge agents serve as foundational tools that gather and synthesize data from multiple sources, such as historical production performance, demand forecasts, inventory levels and resource availability. By harnessing this data, an agent provides planners with actionable insights. It can analyse past trends to identify potential bottlenecks, recommend best practices and highlight opportunities to streamline production processes. Planners can query the agent to retrieve detailed analysis or predictive insights, making it a valuable decision-support tool. Over time, it refines its knowledge, continuously improving the accuracy and relevance of its insights. Adviser agent Parameter adviser agents continuously monitor machine performance, detect real-time deviations and recommend setpoint improvements to achieve desired production goals. Operators can validate the recommended settings and corrective actions. The agent refines its recommendations over time by integrating real-time worker feedback and assessing performance outcomes.Planning adviser agents build on the capabilities of the knowledge agent by continuously monitoring real-time production and forecast data. They detect potential deviations, such as delays, resource shortages or equipment downtime, and recommend proactive adjustments to the production plan. These recommendations can include rescheduling, resource reallocation or inventory adjustments to ensure operational targets are met. The agent learns from the planner’s validation and feedback, allowing it to improve its predictions and better anticipate future challenges. This adaptability augments its ability to optimize production planning. Automation agent Parameter setpoint automation agents continuously track machine performance, identify anomalies in real time and autonomously adjusts setpoints or take corrective actions. They adapt and self-correct parameters based on current production priorities without requiring human intervention. The agent can also work with digital twins, incorporating necessary external inputs such as customer demands into the digital twin to support business decisions. Although pilots in series production are still pending, a research study has demonstrated how LLM agents can control and steer operations remotely, either with human oversight or through AI agents in a digital twin.6Planning automation agents take the next step by autonomously managing production schedules in real time. They respond dynamically to production events like machine breakdowns, labour shortages or fluctuations in demand. The agent continuously updates the production plan, reallocates resources and reschedules tasks to ensure the overall production process remains efficient. Unlike the adviser agent, the automation agent acts independently to adjust, only requiring human oversight for major exceptions or strategic decisions. Meta agent Meta parameter agents oversee and orchestrate multiple dedicated machine parameter agents, synchronizing setpoints across various machines. This agent autonomously coordinates actions to ensure an optimized production flow, preventing bottlenecks and dynamically adjusting the system to optimize overall performance.Meta supply chain agent oversees multiple planning agents that focus on specific areas such as labour, inventory, or capacity. It coordinates these agents to ensure that the entire production process remains balanced and optimized. By dynamically adjusting priorities and resources across various departments, the meta agent ensures that the production plan aligns with broader organizational goals and avoids conflicts between localized planning decisions. Frontier Technologies in Industrial Operations 13

Frontier Technologies in Industrial Operations 2025