• Written by Azhar Aliyar, Director

A production planner sits at her desk, staring at the output report from the manufacturing execution system, trying to understand why the numbers do not match the production schedule she received from the enterprise resource planning system. This discrepancy is a common problem in many organisations, and it highlights the need for careful integration of AI systems with the systems of record in production.

Read paths, write paths and why the second needs more care

The integration of AI systems with production systems involves creating read paths to access the necessary data and write paths to update the systems of record. While read paths are relatively straightforward, write paths require more care to avoid creating a second source of truth. If the AI system updates the production schedule without properly synchronising with the enterprise resource planning system, it can lead to inconsistencies and errors.

A shift supervisor checks the maintenance log to ensure that the equipment is properly maintained, but if the AI system updates the log without notifying the maintenance team, it can cause confusion and delays. Therefore, it is essential to carefully design the write paths to ensure that the AI system updates the systems of record in a controlled and consistent manner.

Reconciling definitions that differ between systems

One of the significant challenges in integrating AI systems with production systems is reconciling definitions that differ between systems. For example, the manufacturing execution system may define a production cycle as the time it takes to produce a batch of products, while the enterprise resource planning system defines it as the time it takes to produce a single unit. These differences in definition can lead to inconsistencies in the data and affect the accuracy of the AI system's predictions.

To overcome this challenge, it is essential to carefully map the definitions and data structures between the different systems. This requires a thorough understanding of the data models and business rules used in each system. By reconciling these differences, organisations can ensure that the AI system receives consistent and accurate data, which is essential for making informed decisions.

The production data this depends on

The integration of AI systems with production systems depends on a range of production data, including output against plan, cycle times, scrap and quality records, and maintenance history. This data is used to train the AI models and make predictions about future production. For example, the AI system may use historical data on output against plan to predict future production levels and identify potential bottlenecks.

The category manager compares two suppliers based on their quality records and delivery times, but if the AI system does not have access to this data, it cannot make informed recommendations. Therefore, it is essential to ensure that the AI system has access to all the necessary production data to make accurate predictions and recommendations.

What it changes about: unplanned downtime and stock cover against the plan

The integration of AI systems with production systems can significantly reduce unplanned downtime and improve stock cover against the plan. By analysing historical data on equipment failures and maintenance schedules, the AI system can predict when equipment is likely to fail and schedule maintenance accordingly. This can help reduce unplanned downtime and ensure that production runs smoothly.

For more information on how AI systems can improve production planning, visit our production page. By integrating AI systems with production systems, organisations can also improve stock cover against the plan by predicting demand and adjusting production levels accordingly. This can help reduce inventory costs and improve customer satisfaction.

Measured with: OEE, unplanned downtime hours, scrap rate

The effectiveness of the integration of AI systems with production systems can be measured using a range of key performance indicators, including overall equipment effectiveness (OEE), unplanned downtime hours, and scrap rate. These metrics provide a clear picture of the production process and help organisations identify areas for improvement.

By monitoring these metrics, organisations can adjust the AI system's parameters and improve its performance over time. For example, if the scrap rate is higher than expected, the organisation may need to adjust the AI system's quality control parameters to improve the accuracy of its predictions.

What this does not do

While the integration of AI systems with production systems can bring significant benefits, it is not a solution for all production problems. It does not replace the need for skilled production staff or address underlying issues with the production process. If the production process is poorly designed or inefficient, the AI system will simply optimise a suboptimal process.

Before implementing an AI system, organisations should ensure that their production process is well-designed and efficient. They should also have a clear understanding of their production data and systems of record. For more information on how to get started with AI systems integration, visit our contact page. By carefully evaluating their production process and systems, organisations can ensure that the integration of AI systems with production systems brings the desired benefits.