• Written by Azhar Aliyar, Director

A quality engineer checking the inspection records for a batch of components will often find that the data is scattered across multiple systems, making it difficult to get a complete picture of the production process. This is a common problem in many organisations, where different systems are used to manage different aspects of quality, such as non-conformance reports, batch genealogy, and customer complaints.

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

When integrating AI systems for quality, it is essential to consider the read paths and write paths. Read paths refer to the flow of data from the systems of record to the AI system, while write paths refer to the flow of data from the AI system back to the systems of record. While read paths are relatively straightforward, write paths require more care, as they can potentially create a second source of truth. This can lead to inconsistencies and errors, which can have serious consequences in a quality control environment.

A shift supervisor checking the maintenance log, for example, may find that the data has been updated in the AI system, but not in the underlying system of record. This can cause confusion and delays, and may even lead to defects or non-conformities going undetected. To avoid this, it is crucial to ensure that write paths are carefully managed, and that data is updated consistently across all systems.

Reconciling definitions that differ between systems

One of the challenges of integrating AI systems for quality is reconciling definitions that differ between systems. For example, one system may define a defect as a non-conformance that requires rework, while another system may define it as a non-conformance that requires scrap. These differences in definition can cause problems when trying to integrate data from multiple systems, as the AI system may not be able to accurately identify and classify defects.

To overcome this challenge, it is necessary to establish a common set of definitions and standards across all systems. This can involve working with subject matter experts to develop a unified vocabulary and set of metrics, and then mapping these definitions to the existing systems. By doing so, the AI system can accurately identify and classify defects, and provide a consistent and reliable picture of quality.

The quality data this depends on

The integration of AI systems for quality depends on a range of quality data, including inspection records, non-conformance reports, batch genealogy, and customer complaints. This data provides the foundation for the AI system to identify patterns and trends, and to make predictions and recommendations. For example, a category manager comparing two suppliers may use data on defects per unit and first pass yield to determine which supplier is more reliable.

Batch genealogy, which tracks the origin and movement of components and materials, is particularly important in this context. By analysing batch genealogy data, the AI system can identify potential sources of defects and non-conformities, and provide recommendations for improvement. Similarly, customer complaints data can provide valuable insights into the root causes of defects and non-conformities, and help to identify areas for improvement.

What it changes about

The integration of AI systems for quality can have a significant impact on the way that defects and non-conformities are traced and addressed. For example, defects that are traced late to a batch or supplier can be quickly identified and addressed, reducing the risk of further defects or non-conformities. Similarly, CAPA (Corrective and Preventive Action) records that are written but never read can be automatically reviewed and updated, ensuring that corrective actions are taken and preventive measures are put in place.

By providing real-time insights and recommendations, the AI system can help to reduce the time to root cause, and improve the overall efficiency and effectiveness of the quality control process. This can be particularly useful in production environments, where defects and non-conformities can have serious consequences for product quality and customer satisfaction.

Measured with

The effectiveness of AI systems integration for quality can be measured using a range of metrics, including first pass yield, defects per unit, and time to root cause. These metrics provide a clear and objective way to evaluate the performance of the AI system, and to identify areas for improvement. By tracking these metrics over time, organisations can see the impact of AI systems integration on their quality control processes, and make adjustments as needed.

For example, a reduction in defects per unit may indicate that the AI system is effectively identifying and addressing potential sources of defects. Similarly, a reduction in time to root cause may indicate that the AI system is providing valuable insights and recommendations for improvement. By using these metrics to evaluate the effectiveness of AI systems integration, organisations can ensure that their quality control processes are operating at optimal levels.

What this does not do

While AI systems integration for quality can provide significant benefits, it is not a replacement for human judgment and expertise. In some cases, the AI system may not be able to accurately identify or classify defects, or may provide recommendations that are not feasible or practical. In these cases, human intervention and oversight are necessary to ensure that the quality control process is operating effectively.

If you are considering AI systems integration for quality, but are unsure about how to proceed, you can contact us for more information and guidance. Our team of experts can help you to evaluate your options, and develop a plan for implementing AI systems integration that meets your specific needs and requirements.