• Written by Azhar Aliyar, Director

In production, systems are supposed to take action, not just answer questions, but in practice, many organisations settle for chatbots that can only respond to a single query, rather than implementing agentic AI workflows that can drive real change. This is a mistake, as agentic AI workflows have the potential to transform production by automating complex processes and improving efficiency.

What separates an agent from a chatbot

An agent is distinct from a chatbot in that it has the ability to use tools, maintain state, and work towards a goal. This means that an agentic AI workflow for production can interact with multiple systems, track progress, and adapt to changing circumstances, rather than simply providing a static response to a query. For example, a production planner can use an agentic AI workflow to automatically adjust the production schedule based on changes in demand or supply chain disruptions.

This is in contrast to a chatbot, which can only respond to a single question or prompt, and lacks the ability to take action or interact with other systems. While chatbots can be useful for providing information or answering simple questions, they are not capable of driving the kind of complex, multi-step processes that are required in production.

Where a human approval step belongs

One of the key benefits of agentic AI workflows is that they can automate many of the repetitive and time-consuming tasks that are currently performed by humans. However, this does not mean that human approval steps are no longer necessary. In fact, removing human approval steps prematurely can be a recipe for disaster, as it can lead to unintended consequences or errors that are not caught until it is too late.

A better approach is to identify where human approval steps are truly necessary, and to implement agentic AI workflows that can automate the surrounding processes. For example, a shift supervisor may need to approve a change to the production schedule, but the agentic AI workflow can automate the process of generating the new schedule and alerting the supervisor to the change.

Production data dependencies

Agentic AI workflows for production depend on a wide range of data sources, including output against plan, cycle times, scrap and quality records, and maintenance history. This data is used to inform the workflow and ensure that it is operating effectively. For example, an agentic AI workflow may use data on output against plan to identify bottlenecks in the production process and adjust the schedule accordingly.

By integrating with existing systems and data sources, agentic AI workflows can provide a more complete and accurate picture of the production process, and enable more effective decision-making. To learn more about how agentic AI workflows can be applied to production, visit our production page.

Impact on unplanned downtime and stock cover

One of the key benefits of agentic AI workflows for production is that they can help to reduce unplanned downtime and improve stock cover against the plan. By automating many of the repetitive and time-consuming tasks that are currently performed by humans, agentic AI workflows can help to identify potential issues before they occur, and take proactive steps to prevent them.

For example, an agentic AI workflow may use data on maintenance history to identify equipment that is at risk of failure, and schedule maintenance accordingly. This can help to reduce unplanned downtime and improve overall efficiency.

Measurement and evaluation

Agentic AI workflows for production can be measured and evaluated using a range of key performance indicators (KPIs), including overall equipment effectiveness (OEE), unplanned downtime hours, and scrap rate. These KPIs provide a clear and objective measure of the workflow's effectiveness, and can be used to identify areas for improvement.

By tracking these KPIs over time, organisations can see the real benefits of implementing agentic AI workflows for production, and make data-driven decisions about how to optimise their operations. If you have any questions about how to get started with agentic AI workflows, please contact us.

Limitations and exceptions

While agentic AI workflows have the potential to transform production, they are not a silver bullet, and there are certain situations where they may not be the best choice. For example, in situations where the production process is highly variable or unpredictable, an agentic AI workflow may struggle to keep up, and human intervention may be necessary.

In these cases, a more traditional approach may be more effective, at least until the process can be stabilised and optimised. Ultimately, the decision to implement an agentic AI workflow will depend on the specific needs and circumstances of the organisation, and should be made on a case-by-case basis.