Current Issue
Material Handling Wholesaler Cover
November 2018
Brian Neuwirth explains how the warehouse of the future may look different from today.

Industry News

View Material Handling Wholesaler's profile on LinkedIn



Improve supply chain management with AI technologies
Artificial intelligence (AI) can provide data to supply chain management (SCM) with predictive analytics and Big Data to help spot potential challenges and issues before they happen. Two steps for using AI and four potential problems are highlighted
Jonathan Wilkins, EU Automation
Jonathan Wilkins, marketing director, EU Automation
Jonathan Wilkins, marketing director, EU Automation

On average, procure-to-pay businesses estimate they spend 55 hours per week doing manual processes and checks, 39 hours chasing invoices, and 23 hours responding to supplier inquiries related to these sources of friction. It's the larger companies-those with over 1,000 employees-that feel this friction the most, according to Tungsten Network. This can be avoided with artificial intelligence (AI) technologies such as distribution software and inventory optimization, which have potential to improve supply chain management (SCM).

The manual, paper-based processes are trivial, but necessary. They equate to 6,500 human hours in the working year according to Tungsten's survey of 442 respondents. Some of this time could be saved by implementing AI and automation. However, these tasks represent just one part of the complex supply chain. Could wider challenges, such as logistics and distribution, also benefit from the AI treatment?

Using AI and Big Data to make better decisions

With volumes of data growing at an unprecedented rate, computers are capable of parsing data in a contextual manner and providing useful insights to an operator without them doing any of the legwork. Big Data technologies also are capable of analyzing market trends, integrating with enterprise systems, and triggering automated actions based on the data collected.

Consider this example: An order has been placed at a steel fasteners warehouse. Automated software allows the system automatically to detect if the product is in stock. If so, automated guided vehicles (AGVs) can be deployed to the exact location in the warehouse to acquire the product for distribution. If the product is not in stock, automation can scan the enterprise resource planning (ERP) system to place a reorder and inform the customer of the estimated dispatch time.

This example shows AI's ability to streamline information so supply chain and logistics functions deliver a competitive advantage to adopters.

In 2017, one California-based telecom equipment manufacturer saw a significant revenue drop from $870 million to $740 million. In an effort to turn things around, the company deployed AI to make better predictions about delivery dates and to analyze past variability and production lead times. The business has since seen revenue pick up again and attributes this to the AI implementation.

However, it's not just large companies that have access to this technology. An increasing number of tech vendors are releasing products so AI technology is accessible at a competitive price. This means AI is fast becoming mainstream, outcompeting traditional supply chain management to enhance everyday business. Companies can begin to reap the rewards of these technologies by following two steps:

Step 1: Build AI readiness

Businesses must have large data sets of deep granularity for AI to be effective. Granularity is used to characterize the scale or level of detail in a set of data, which AI is highly dependent on. The greater the granularity, the deeper the level of detail across the data. Whether AI implementation is in the forthcoming plans or not, it's a good idea to ensure data collection and storage are geared for high granularity.

Boosting granularity may mean increasing the frequency of data readings, refining the precision of such recordings, or even placing sensors in new places to measure new variables. For example, if a flow meter currently is measuring the flow rate of a liquid in litres per minute, changing this recording to milliliters per minute may provide more insightful data. Ultimately, even if a business is not AI-ready today, improving granularity and data collection and analysis will lay the foundation for when AI inevitably becomes a more widespread competitive differentiator. 

Step 2: Target a specific problem

Have one business goal in mind at the beginning. Focusing efforts and resources on a single problem means a significant pain point can be tackled effectively, with relatively low risk compared to a complete overhaul of processes. By selecting a discrete project, initial successes can be built upon and lessons can be learned. These lessons can then be applied to other areas in the supply chain with time.

Here are four targets for companies to consider:

1. Reduce late payments

Client payments and their associated enquiries are huge causes of friction in the supply chain. Using AI to predict which customers are likely to pay late means alternative plans can be put in place. Perhaps these customers need automated chaser e-mails. Maybe it means buyers won't purchase six months of stock for this customer-just in case they are unable to continue with the deal. 

2. Accurately plan equipment supply

Supply chain planning, which involves using intelligent work tools to build concrete plans for things that could go wrong, is a crucial activity. For example, using data from past equipment and current machine performance allows AI to accurately predict when a part will need to be replaced. This is critical, particularly for obsolete legacy equipment, as lead times to find the part from an obsolete parts supplier will vary. 

3. Predict demand for services and products

The demand for services or products can be predicted accurately using market trends and previous sales data. This means price points can be adjusted accordingly to maximize profits. It doesn't stop there. Having insight into future demand also has huge effects on other parts of a business, such as inventory management. 

4. Forecast inventory needs

With this insight on future demand, AI can also help with forecasting inventory based on previous delivery information. This means decisions can be made to optimize stock levels. For example, if AI lets the user know many other businesses will want the same equipment in 12 months' time, this gives the user the opportunity to get ahead by ordering it much sooner than this.

Why does it matter if inventory levels aren't optimized? In 2015, the cost to companies of over-stocking was $470 billion, and under-stocking was $630 billion worldwide, according to IHL Group. Freeing up cash and storage space optimally creates savings.

As the complex web of production and distribution are opened up to the benefits of AI, the supply chain will have a bigger economic impact than any other technology application. McKinsey & Company, a management consulting company, estimates firms will derive between $1.3 billion and $2.1 billion a year in economic value by implementing AI into the supply chain.

However, for businesses starting out with this technology, the focus should remain on building data granularity and choosing a specific issue to overcome with this technology.

Jonathan Wilkins, marketing director, EU Automation. Edited by Chris Vavra, production editor, Control Engineering, CFE Media.
-End-  


ADVERTISEMENTS