• Written by Azhar Aliyar, Director

When a production planner checks the schedule for the day's print jobs, they expect to see a list of orders with their respective quantities, substrates, and due dates. However, if the data in the production planning system is incomplete or inaccurate, the planner may end up with a schedule that is impossible to fulfill, leading to setup and makeready losses, quoting inaccuracies, and substrate waste. This is known as a data foundation failure, and it is a major obstacle to implementing AI-powered query systems like AsscherAi in printing and packaging organisations.

Printing and packaging complexities

The printing and packaging industry presents unique challenges when it comes to data foundation. The high setup costs and job-shop economics mean that even small errors in data can result in significant losses. For example, if the data on substrate availability is incorrect, the production planner may schedule a job that cannot be fulfilled, resulting in wasted time and resources. Similarly, inaccurate quoting data can lead to lost sales or damaged customer relationships.

The complexity of printing and packaging workflows also makes it harder to ensure data accuracy. With multiple machines, substrates, and processes involved, there are many opportunities for errors to creep in. Furthermore, the industry's emphasis on just-in-time production and tight deadlines means that there is little room for error or delay.

Field-level meaning and documentation

One of the main causes of data foundation failures is the lack of clear field-level meaning and documentation. Column names in a database are not sufficient documentation, as they often do not provide enough context for users to understand what the data represents. For example, a column named "quantity" could refer to the quantity of substrate, the quantity of finished products, or something else entirely.

To avoid this problem, it is essential to have clear and concise documentation of each field in the database, including its meaning, format, and any relevant business rules. This documentation should be easily accessible to all users and should be updated regularly to reflect changes in the business or the data.

Out-of-scope systems and partial answers

Another common issue in printing and packaging organisations is the tendency to leave certain systems or datasets out of scope when implementing AI-powered query systems. This can result in partial answers that do not take into account all the relevant data. For example, if the system does not have access to data from the accounting system, it may not be able to provide accurate answers to questions about costs or profitability.

This problem can be particularly acute in organisations with complex workflows or multiple sites. In these cases, it is essential to have a comprehensive understanding of all the systems and datasets involved and to ensure that they are all integrated into the AI-powered query system. The AsscherAi platform can help with this by providing a unified view of all the organisation's data.

Stockouts, backfills, and other gaps

Stockouts, backfills, and other gaps in the data can also teach a model the wrong lesson and lead to inaccurate answers. For example, if the data on substrate availability is missing for a particular period, the model may assume that the substrate was always available during that time, when in fact it was not. This can result in inaccurate forecasts and production plans.

To avoid this problem, it is essential to have a complete and accurate dataset that reflects all the relevant information. This may involve filling in gaps in the data or using alternative data sources to supplement the existing data. The AsscherAi support team can help with this by providing guidance on data preparation and integration.

Limitations and wrong choices

While AI-powered query systems like AsscherAi can be very powerful tools for printing and packaging organisations, they are not a panacea for all data-related problems. In some cases, the data foundation may be so weak that it is not possible to implement an AI-powered query system at all. In other cases, the organisation may not have the necessary resources or expertise to implement and maintain the system.

In these cases, it may be better to focus on building a strong data foundation before attempting to implement an AI-powered query system. This can involve investing in data governance, data quality, and data integration initiatives, as well as providing training and support for users. By taking a step-by-step approach to building a strong data foundation, organisations can ensure that they get the most out of their AI-powered query system and avoid common pitfalls like data foundation failures.