The value of AI in manufacturing lies in reducing product defects, automating quality control, increasing capacity and streamlining maintenance. Highly automated manufactures are ideal candidates for sophisticated analytics as they are already well-instrumented with sensors and controllers. However, less automated manufactures can also beneficially utilize robotics and AI by eliminating inefficient manual processes.

According to a global consulting firm McKinsey, manufacturing industries have captured about only 20%-30% of the potential value of data and analytics to date – and most of that has occurred at a handful of industry-leading companies. This is despite the fact that advanced manufacturers have been working with automation and lean ways of working for decades. By embarking on an AI journey, far more complicated and bigger amounts of information can be processed across manufacturing operations and across the supply chain – detecting inefficiencies, relationships and predicting possible issues before they even emerge.

Many applications are using different regression and classification models and even natural language processing.

Areas of AI application in Manufacturing

  • Quality control can be enhanced through vision systems by adding image recognition models based on convolutional neural networks e.g. to inspect and sort for defects.
  • Downtime of equipment can be minimized by analyzing data for each manufacturing step and/or for a chain of linked operations e.g. moldings importance on assembly output.
  • Predictive maintenance can proactively identify problems related to spare-parts, software, hardware and firmware to eliminate possible points of failure or degraded performance.
  • In general: models can be implemented so that they continuously learn from data, accounting for changing conditions and the current state of a specific machine.

Use cases

In-process controls – errors in Manufacturing
Many product errors or machine malfunctions are spotted late in the manufacturing process due to manual in-process controls
Production planning
A model can identify hidden patterns in historical sales data, such as those in seasonal demand and subsidiary behavior or new product launches
Increase output – perfect batch
Modelling patterns between input and output of the batches as well as performance parameters analysis can increase output substantially
Prediction of service cases in Manufacturing
A supervised algorithm predicts service cases, increasing accuracy in planning of service and maintenance
Sorting and scrap reduction in Manufacturing
Scrap of good products should be minimized and scrap of poor products should be close to hundred percent.
Predictive maintenance in Manufacturing
An algorithm predicts the risk of equipment failure in production
Inventory prediction in Manufacturing
The sale of goods gets predicted and based on that algorithm gives recommendations for store inventory management, allowing inventory or trash reduction
Document search in Manufacturing
Searching and classifying quality documents can save time and help identify areas for improvement

Meet the industry leader

Kim Tosti

Kim Tosti

Manufacturing, 2021.AI

Kim Tosti is the Head of Pharma and Biotech at 2021.AI. His goal is to bring AI & ML onto the agenda of the Pharma industry. Kim’s core competencies include corporate governance, supply chain excellence, and risk management. Prior to 2021.AI, Kim spent the majority of his career at Novo Nordisk as a Senior VP. in

Get in touch

Interested in taking AI into production and scale to every corner of your organization?

Book a demo