MHI Solutions Improving Supply Chain Performance
AI technologies ranked highest of innovations that respondents to the 2024 MHI Annual Industry Report survey expected to adopt within the next five years. In addition to the 58% who expect to adopt AI technology, 27% of respondents report already using AI and only 15% say that they are unlikely to adopt AI tools.
Leveraging the power of AI is a big topic right now, often one of the top five things to do according to many surveys of supply chain leaders, said Arturo Buzzalino, vice president of product management for MHI member Epicor. Supply chain companies are using enterprise resource planning (ERP) technologies to format data, display financial information and report on inventory, but the addition of AI can enhance analysis and free up staff time to focus on higher-level tasks.
Because AI is dependent on data, a robust ERP is required. “If a company is using an old system or still relying on spreadsheets to track business operations, it is not ready for AI,” said Buzzalino. Before exploring adoption of AI, evaluate the quality of data in the existing ERP and ensure that it accurately reflects the reality of the business.”
“Whenever a company adopts a new tool—and AI is a tool—the process of adoption is important,” said Divya Prakash, director of business consulting, Industry 4.0, at MHI member SICK. “No one should implement a new technology because it is the latest buzzword.” AI is a strong tool with many benefits if the decision on when to adopt and how to implement is carefully made, he added.
Set Goals; Assess Readiness
“The first step is to set up a strategy team that includes leaders from all areas of the business—operations, maintenance, IT, finance and sales,” said Prakash. “This team should determine what the goals are for AI implementation across the organization.”
The strategy team should ask questions that include: How will we use AI? What problems can the technology solve? and, What outcomes do we want to see? suggested Prakash. The answers should be very specific and reflect organizational goals such as improving performance, reducing costs or enhancing worker safety, he added.
“The next step is to conduct a readiness assessment to identify gaps in infrastructure that need to be addressed before implementing AI tools,” said Prakash. For example, producing dashboards with real-time information on inventory or orders to make it convenient for employees working on the floor of a distribution center are not effective if there is no connectivity to provide real-time data to devices.
“Data quality also needs to be assessed and processes need to be standardized to improve accuracy of data that is used by AI tools,” said Prakash. “Data is the lifeblood of AI so be sure that you have a robust data pipeline along with storage space and a strong security and compliance program in place.” This is a critical step when preparing to implement AI tools because AI models rely on volumes of data to learn how best to analyze information to present an accurate picture of business activities or to make recommendations and predictions. “You want to ensure that the business data as well as customer information is protected and used appropriately.”
According to one industry report, AI performance improves as organizations move through the levels of data maturity.2 The levels of data maturity and the percentage of supply chain companies at each level are:
- Stage 1: Six percent of supply chain companies have undefined data management processes and zero or limited use of data analytics.
- Stage 2: Data management policy is driven at the local level without alignment with the global organization in 44% of the supply chain.
- Stage 3: Global organization drives the standardized data management policy in 35% of organizations.
- Stage 4: Twelve percent of supply chain companies have standardized data management that is measured across the organization.
- Stage 5: Only 3% of the supply is at the stage where data management is standardized, measured and continuously improved with a focus on learning.
Start with a Pilot Project
After looking at the potential uses of AI that make sense in the business, the strategy team needs to zoom in on one area or issue to identify a pilot project, said Prakash. The pilot project should be well-defined and should provide practical experience on the use and benefits of AI. Setting goals, identifying expected outcomes and defining a timeline for evaluation will help avoid two pitfalls of introducing any new technology or process. “Don’t get stuck in ‘analysis paralysis’ in the pilot project,” he said. “Once you have enough information to evaluate AI’s effectiveness, make the decision and move forward to use it for other areas or not to implement.”
When expanding the use of AI in the organization, don’t forget to set clear objectives, said Prakash. “If goals are ambiguous, you are at risk for scope creep,” he explained.
Expanding the parameters of the pilot project or allowing the implementation of AI by disparate departments without an overall plan can result in less-than-optimal results, which frustrates users and reduces the effectiveness of AI. “The most overlooked aspect of AI implementation is ensuring that the AI model has been built with good data that produces an unbiased result,” said Prakash. “It’s also important that implementation be coordinated to ensure that AI is integrated into the existing system in a way that minimizes disruption to workflow.”
Click here to read the full feature.