• Written by Azhar Aliyar, Director

A production planner at a tier one supplier is deciding this week how to allocate production capacity to meet the demands of multiple automotive original equipment manufacturers, and is currently basing it on manual checks of the enterprise resource planning system and phone calls to the OEMs' logistics teams. This process is prone to errors and delays, and the planner is looking for a way to integrate the supplier's systems with the OEMs' systems to get a more accurate and up-to-date view of demand.

Automotive industry challenges

The automotive industry presents several challenges that make integration harder, including supplier escalations, engineering change propagation, and warranty root cause analysis. Supplier escalations occur when a supplier is unable to meet the demand of an OEM, and the OEM must find an alternative supplier or risk production delays. Engineering change propagation refers to the process of updating designs and specifications across multiple systems and stakeholders, which can be time-consuming and error-prone. Warranty root cause analysis involves identifying the underlying cause of a warranty claim, which can require access to data from multiple systems, including production, quality control, and customer service.

These challenges are exacerbated by the tiered supply chain and tight takt times in the automotive industry, which means that even small delays or errors can have significant consequences. For example, a delay in receiving parts from a supplier can cause a production line to shut down, resulting in lost production and revenue.

Read and write paths

When integrating AI systems with the systems of record in the automotive industry, it is essential to consider both read paths and write paths. Read paths refer to the process of retrieving data from the systems of record, while write paths refer to the process of updating the systems of record with new data. While read paths are relatively straightforward, write paths require more care, as they can potentially introduce errors or inconsistencies into the systems of record.

A good example of this is when a shift supervisor checks the maintenance log to see if a machine is due for maintenance, and then updates the log to reflect that the maintenance has been performed. In this case, the read path is simple, but the write path requires careful consideration to ensure that the update is accurate and consistent with other data in the system.

Reconciling definitions

Another challenge in integrating AI systems with the systems of record in the automotive industry is reconciling definitions that differ between systems. For example, one system may define a "vehicle" as a single unit, while another system may define it as a collection of parts and components. These differences in definition can make it difficult to integrate data from multiple systems, and can lead to errors or inconsistencies if not properly addressed.

To reconcile these differences, it is essential to establish a common set of definitions and standards that can be used across all systems. This can involve working with stakeholders to agree on a common vocabulary and set of definitions, and then implementing these definitions consistently across all systems. For more information on how to establish a common set of definitions and standards, please visit our website.

Latency budgets and precomputation

When integrating AI systems with the systems of record in the automotive industry, it is also essential to consider latency budgets and precomputation. Latency budgets refer to the maximum amount of time that can be spent on a particular task or process, while precomputation refers to the process of computing and storing data in advance to reduce latency. By carefully managing latency budgets and precomputing data where possible, it is possible to improve the performance and responsiveness of AI systems.

For example, a category manager may want to compare the prices of two different suppliers, but the data may be stored in a database that is only updated once a day. In this case, precomputing the data and storing it in a cache can improve the responsiveness of the system and allow the category manager to make more informed decisions. If you have any questions about how to implement precomputation in your AI system, please contact us.

Limitations and wrong choices

Finally, it is essential to recognize the limitations of AI systems integration in the automotive industry, and to understand when it is the wrong choice. AI systems integration is not a solution for every problem, and it is not suitable for every organization. For example, if an organization has a simple and well-defined process, it may not be necessary to integrate AI systems. Similarly, if an organization has a high degree of variability in its processes, it may be difficult to integrate AI systems effectively.

In these cases, it may be better to focus on simplifying and standardizing processes, rather than trying to integrate AI systems. By understanding the limitations and challenges of AI systems integration, organizations can make more informed decisions about when and how to use these technologies. Some of the key considerations include the complexity of the process, the availability of data, and the potential benefits of integration.

For instance, the decision to integrate AI systems should be based on a careful analysis of the potential benefits and challenges, rather than on a desire to adopt new technology. By taking a thoughtful and informed approach to AI systems integration, organizations can avoid common pitfalls and achieve greater success in their integration efforts.