Blog

How AI and Automation Can Support Product Compliance Across the Supply Chain

Product quality is a key focus of any manufacturing company, and issues can make or break the reputation of a brand. In many cases, though, the quality management process can be complex and under-resourced, with issues turning into costly problems before they are detected and resolved.

Companies all along the supply chain are collecting data about the products they provide, but in most cases, there are limited analyses and insights coming from that data. Each company has its own method and format for collecting and storing that data, and it is simply too time-consuming to process that data manually to gain actionable insights.

Technologies such as artificial intelligence, machine learning, and natural language processing enable companies to achieve the speed and insight that scale to improve quality, ensure compliance, and promote continuous improvement, removing data silos and enabling end-to-end analysis.

Complex Transparency Requirements from Regulators, Consumers, and Retailers

Product requirements for manufacturers are constantly increasing and becoming more complex. Those requirements are not only coming from regulatory bodies, but also from consumers and retailers demanding greater transparency as to what is in their product, as well as from where that product and its components come from.

Multinational consumer goods company Unilever, for example, has spent the past several years transforming its quality management processes to adapt to those heightened requirements, as well as the increasing number of channels by which it does business. Ahmed Maklad, Global Digital Quality Transformation Director with Unilever, discussed their transformation and Unilever’s partnership with Veeva and Amazon Web Services at Veeva’s 2021 Quality & Regulatory Global Summit.

Watch How AI and Automation Can Support Product Compliance Across Your Supply  Chain on demand.

“Consumers are looking for more transparent, trusted, and purposeful brands, and are also looking for products that are more efficient in terms of value and functionality,” Maklad said. “Markets now have more signals, faster innovations, and new channels, and this, of course, led us to move through a path of [digital] transformation.”

Veeva’s Partnership with AWS & Unilever to Automate CoAs

Unilever tried to address two main problems over the course of its digital transformation, Maklad said. The first was the question of how to drive the business of today and tomorrow with more speed, experimentation, and cost efficiency while continuing to provide consumer safety and maintaining brand dedication. The second was to determine how the quality of the consumer experience could help drive consumer choice.

The starting point for discussions with Veeva about addressing those problems focused on certificates of analysis (COAs) and how to digitize Unilever’s workflows. A COA is a document that Unilever’s supply partners send to the company’s factories, either with raw material or packaged material, outlining the product’s specifications. The quality management team at the factory analyzes the material, compares it to Unilever’s specifications, and records that analysis along with the COA in the company’s internal data quality management systems.

Before the partnership with Veeva, the recording process had been a completely manual operation, taking between five and 15 minutes per shipment. And with millions of COAs coming into factories around the world over the course of a year, the overall time investment was staggering.

Data Quality Management

Unilever realized that by leveraging AI in quality management and digitizing that operation, it could not only save time but also move its quality management team from a focus on inspection to a focus on prevention and improvement, making smarter decisions along the way.

“We started by thinking about how to use OCR technology and machine learning on scanning the COAs,” Maklad said.

“Our supply partners will upload their information to the Veeva QualityOne platform, and we use machine learning to compare that to the specifications,” he said. “That information is recorded and can even be used to determine whether a shipment is accepted or rejected.”

Veeva QualityOne leverages the Amazon Textract machine learning service to extract text, handwriting, and data from scanned documents as part of an overall automated quality management system. QualityOne manages quality processes across all stakeholders, including external suppliers, offering access and collaboration from anywhere around the globe.

The Data Quality Problem

One of the challenges in digitizing the COA workflow was the fact that suppliers often use different terminology for the same concept. Consider, for example, the term “moisture content.” One supplier might just say water, while another might say water content, and another might say moisture content.

“To be able to use machine learning and be able to extract that information without having templates was the challenge,” said Raj Coppaparu, Technical Product Management Leader with AWS. “What we did was partner closely with Veeva and Unilever to be able to train Amazon Textract specifically on those types of documents in order to improve accuracy.”

Machine Learning and AI in Quality Management

Through its partnership with Veeva and AWS, Unilever was able to accomplish three goals: increase the speed at which COAs were processed, capture more and better information from those documents, and cost savings due to increased process efficiency. The ability to extract and process data quickly and accurately was central to those efforts.

The Bottom Line

Through COA automation, Unilever is digitizing data, removing silos, and enabling end-to-end analysis. 

Unilever’s journey has provided it with more and better-structured data that offers a clear view of pain points, allowing it to focus on problems and fix them before they turn into crises.

“Data is one of the biggest elements that would define the destination and the speed of this transformation journey,” Maklad said. “The data, the cleanness of the data, the more data structure you have will determine if this journey of transformation will be an easy one or a hard one.”

Learn more about how Veeva can help manage that transformation.

Sophia Finn

Posted by Sophia Finn

Sophia Finn manages the customer success program. Prior to her adventure at Veeva, she worked for the FDA as a microbiologist testing flu vaccine efficacy and food samples for bacterial contamination. She also worked for various medical device manufactures where she remediated quality processes, and managed CAPA, complaints/MDR, and deviation programs. Prior to all of this, Sophia served 7 years in the Air Force as a medic, and we thank her for her service.