AI in Supply Chain Management

General Mills has faced no shortage of supply chain disruptions in recent years: commodity inflation, war in one of the world’s largest grain-producing regions, labor shortages, freight and port snarls, and all the various effects of the pandemic.

But the packaged food giant was already building resilience into its supply chain before those events. “The nature of supply chain is always you're dealing with complexities,” Paul Gallagher, General Mills’ chief supply chain officer, said in an interview. “You're always dealing with some kind of disruption, whether that's disruption that you create from within in order to drive a different proposition, or disruption that's external.”

In other words, stuff happens. But resilient supply chains can react quickly.

Gallagher joined the company in 2019 as VP of its North America supply chain, where he was tasked with transforming its supply chain network to drive competitive advantage for the business.

He says the company began digitizing before the pandemic and doubled down during the COVID-19 crisis “to ensure that we were driving value and visibility to all the disruptions.”

The company’s effort to digitally connect its supply chain starts with sourcing. The resulting data of that linking “give[s] us visibility of potential risk and opportunity, whether that be weather patterns, or whether that be supplier disruptions or demand on suppliers, or what happened recently with the Suez Canal,” Gallagher said.

He added, “We now are so connected in understanding that it's not requiring an individual to go off and find the information — the information is already available, and it's prompting us with optionality around what to do.”

To achieve end-to-end visibility and analysis, the company has built a digital twin of its supply chain that runs on an advanced analytics and artificial intelligence platform, which General Mills is building with software firm Palantir.

“It's new to us, especially in the in the food industry,” Gallagher said. “Principally, it allows us to take multiple inputs and insights every day on that intelligent execution platform, and it consumes real-time data.”

Looking at just the company’s North American dry foods business, the system takes over 3,000 orders, assesses them overnight, automatically makes recommendations to an “already optimized plan,” and then produces 500 recommendations depending on variable inputs. All of that digital work helps automate certain supply chain planning functions and helps teams “make more learned and insightful decisions and trade-offs,” Gallagher said.

Along with visibility, these investments hold the potential to ramp up productivity. “What used to take a day to do is now taking us minutes,” Gallagher said. With those gains, General Mills is eyeing expanded margins. It also aims to free up supply chain staff to work more closely with customers on longer-term opportunities and so they are not “caught in the tyranny of the urgent,” Gallagher said.

As well as cutting back costs, visibility and analytics can help General Mills reduce its overall climate and sustainability footprint.

“We always had the ability to track” emissions in the supply chain, Gallagher noted. “But it required somebody to go in and assess it all. We now see it as it happens, and it's captured as it happens. So we've moved from episodic to actually dynamic.”

MANIFEST VEGAS 2024 — Schneider Electric leveraged artificial intelligence and machine learning technology to eliminate maintenance issues at its automated smart factory in Kentucky, Kenneth Engel, SVP of internet giants at the company, said at the Manifest Vegas 2024 conference.

Schneider had put in a mile-long overhead conveyor system above the manufacturing operations to replace trucks that were previously used inside the factory. The new system has 1,700 chain links and hooks that move the products and materials from one place to the next, and is equipped with radio frequency identification technology to help accurately move several thousand SKUs.

“Jams happen because, as you might imagine, the factory has several transfer points where there’s handoffs from one hook to another hook,” Engel said. “So, the biggest problem that we had was to figure out how to stop the chain breaks and be more predictive with that.”

The company used AI and machine learning to determine where to put sensors in an overhead conveyor system, partly by analyzing the system’s digital twin.

The digital automation and energy management specialist used its company-owned Aveva software to notify an employee when there might be a problem or if the chain breaks, Engel noted.

“And it’s working,” he added. “We’ve eliminated the problem completely through the use of AI and machine learning.”

Schneider Electric’s 5,000 square-foot automated Lexington, Kentucky, factory takes in sheets of foil and transforms them into products. More specifically, the site manufactures load centers with circuit breakers and commercial safety switches, Engel said.

Products also move from the smart factory straight to the warehouse and are ordered on a “real time basis.” Schneider used technologies like AI and machine learning to enhance efficiency in its operations. In turn, this example has saved the company hundreds of thousands of dollars of labor in addition to time and delivery performance to Schneider’s customers, Engel said.

C.H. Robinson Worldwide is incrementally turning tasks over to artificial intelligence technology, as the freight brokerage giant seeks to bolster productivity.

The company has announced various efforts to automate logistics management processes, such as responding to emailed quote requests from shippers and touchless appointments. These moves aim to free up time for its shrinking workforce to focus on customer retention and relationship building as well as revenue growth.   

C.H. Robinson’s strategy appears to be paying off, according to executives on the broker's Q1 earnings call. President and CEO Dave Bozeman — who was hired in 2023 to improve efficiency — said he was encouraged by the company's productivity improvements. Later on the call, COO Arjun Rajan credited the company's use of generative AI in daily tasks as the source of the new efficiencies.

Rajan said concurrent work streams helped eliminate productivity bottlenecks in Q1, improving customer and carrier experiences.

"One of those work streams is aimed at using generative AI to automatically respond to transactional truckload quote emails to drive faster speed to market, increase our addressable demand, and reduce manual touches," he said. 

How GenAI is being used for email quotes

C.H. Robinson sees emailed requests for quotes as an important use case for generative AI. Automating the process can eliminate human involvement in mundane tasks, letting staff focus on higher-return actions such as upselling and troubleshooting.

"Responding to transactional truckload quote requests is time-consuming for account teams and we must respond quickly to be competitive," Rajan said. "Through our automated process and utilizing our GenAI technology, more than 2,000 truckload customers are getting the benefit of faster response time with our automated email quotes, and we will continue to scale this technology to cover more customers and other modes."

The company started by developing proprietary technology that uses artificial intelligence to classify and read incoming emails. AI then determines if the inquiry was seeking a quote, and if so, will reply with pricing.

The company said its automated process can provide a quote in just over 2 minutes. The system replies to 2,000 customer quotes daily, C.H. Robinson said, noting the achievement opens doors to automating other transactions shippers and carriers choose to do by email.

While the company has other technology and applications that can provide instant price quotes, the AI-enabled solution is a breakthrough for customers who prefer email, Mark Albrecht, VP for artificial intelligence, said in the announcement.

“Email works the same for everybody,” he said. “There are no fields to fill in.”

Touchless appointments leverage a similar solution 

Months before C.H. Robinson rolled out automated email quotes, the company released another productivity-boosting solution: touchless appointments.

“It’s far more efficient for technology to find an appointment slot that’s open, that works for both the loading dock and the carrier and gets the freight where it needs to be on time,” Michael Castagnetto, president of North America surface transportation, said in the announcement.

The two solutions have several similarities. First, C.H. Robinson is marketing them for their ability to free up staff's time for other, more productive tasks. But second, they both rely on AI to solve one of logistics' lingering technical challenges — reading and responding to emails.

Noah Hoffman, vice president of North America surface transportation, explained the importance of the technological breakthrough in a March interview.

Hoffman said when he began working at C.H. Robinson about 20 years ago, his desk had two telephones — one for taking shipments and the other for contacting receiving points. While a lot has changed since then, the appointment-scheduling process is often still labor-intensive, requiring clerks to seek out load details like cargo pick-up times, ready dates or labor availability, then match them with an ideal appointment time on various technological systems.

AI has enabled C.H. Robinson to automate several steps of that process, such as instantly matching appointment availability to book an ideal time, Hoffman said. Just as importantly, the technology helps even when shippers prefer to exchange details via email — as AI is able to process emails for key information, and automate C.H. Robinson's side of the process.

The result, Hoffman said, has been higher service for the customers using the technology which, as of March, were predominantly in the food and beverage space.

“Leveraging our scale with our data and then empowering an automated process, which leverages artificial intelligence to find optimal appointments, drives a higher-level service from both on-time pick up and on-time delivery,” Hoffman said.

Service, productivity, and leveraging scale for competitive advantage were also themes executives on C.H. Robinson's earnings calls highlighted when discussing technology.

Technology will be key to achieving the company's goals of boosting North America Surface Transportation shipments by 15% per person per day and growing Global Forwarding shipments by 10% per person per month, C.H. Robinson's COO said on the earnings call.

"GenAI puts the power of large language models into the hands of our frontline teams," Rajan said. "With more data and history to leverage than any other 3PL, we have opportunities to harness the power that Generative AI now offers to further capitalize on our information advantage and we'll continue to look for and pursue those opportunities."

Dive Brief:

• FedEx has launched "a much better machine-learning model" to provide more accurate estimates to customers about delivery times, President and CEO Raj Subramaniam said at the CNBC Evolve Global Summit.
• The model factors in shipping variables such as weather patterns and traffic conditions. "We have built deep learning models [that] recognize these patterns and they get better every single day, and the predictability gets better," Subramaniam said.
• The company is also exploring the capabilities of generative AI technologies that power chatbots like ChatGPT. These models can improve the delivery experience by asking users questions about their shipment and "predict for you what harmonized code that we need to provide," Subramaniam said.

Dive Insight:

FedEx has tapped into machine learning for the past couple of years to improve estimated delivery time accuracy, and its precision is expected to improve with continued use.

The company's interest in machine learning underscores its transformation over the past few years to become what Subramaniam called "a data-driven, digital-first company." FedEx is particularly interested in leveraging the information it gathers from the 16 million packages it delivers daily.

"When you see a FedEx truck on the street, you see a truck with FedEx packages," Subramaniam said. "I see logistics intelligence on that truck."

Harnessing the power of machine learning can help FedEx maximize the value of this data. The logistics giant is currently using machine learning to develop more accurate volume forecasts in its Ground unit and provide shippers with predictive carbon emissions data through its FedEx Sustainability Insights platform, executives said on a June earnings call.

In terms of using AI-powered language models, Subramaniam said FedEx has "a head start,” with its data organized on the right platform to make the most out of the technology.

"We need to have data platforms that these models can run on so that we can create insights," Subramaniam said.

Rival UPS is also tapping into emerging technological tools to strengthen its operations. It adjusted the flow of packages earlier this year via machine learning in response to shrinking demand. CEO Carol Tomé said in August that UPS' network planning tools are now able to “do in an afternoon what used to take a team of engineers months to do."

Dive Brief:

• Aldi International Buying Asia, the grocery chain’s global hub to source products from Asia, is turning to artificial intelligence as a way to strengthen its end-to-end freight management, according to a Jan. 4 press release.
• The supermarket chain aims to centralize its global shipping volume, increase cost transparency and improve its control over the movement of goods throughout its supply chain using One Network Enterprises’ Intelligent Control Tower on the Digital Supply Chain Network solution.
• The technology platform “will greatly improve effectiveness in collaborating with our suppliers and being more responsive to our customers,” Fritz Walleczek, managing director of Aldi International Buying Asia, said in a statement.

Dive Insight:

Besides being a sourcing hub, Aldi’s Hong Kong-based International Buying Asia unit manages all global ocean freight logistics for the grocer. The control tower technology offers Aldi a way to build a “strategic ocean freight procurement strategy to manage their global shipments of goods better,” according to One Network Enterprises’ press release.

Aldi also aims to use the technology to specifically consolidate and reduce freight rates, shipment handling and storage costs as well as manual processes.

The solution is designed to match demand with available supply in real time by integrating Aldi’s network of logistics partners and suppliers around a single data model. This will “enable real-time collaboration across trading partners, workflows, and applications,” according to the release.

As more companies look to gain greater real-time visibility and collaboration across its supply chain, more have been looking toward AI and other technology platforms to mitigate delays and other issues spurred by changing market conditions.

Allbirds, for instance, recently implemented Anaplan’s planning software to streamline its demand, inventory and supply planning processes. Sports equipment retailer Brooks Running also decided to implement a tech platform in 2023 to deepen its supply chain visibility in a bid to achieve its responsible sourcing goals.

Dive Brief:

• Coca-Cola signed a $1.1 billion five-year deal with Microsoft to use the technology company’s cloud computing and artificial intelligence services.
• Coca-Cola has been innovating with generative AI for nearly a year and has turned to Microsoft’s Azure OpenAI Service for its supply chain, manufacturing and marketing. The beverage giant said it is exploring the use of AI to improve customer experiences, streamline operations, foster innovation, improve its competitive advantage, boost efficiency and discover growth opportunities.
• Coca-Cola previously signed a five-year deal in 2020 for $250 million to use Microsoft's cloud and business software.

Dive Insight:

As food and beverage manufacturers incorporate artificial intelligence into more facets of their operations to give them an advantage in the marketplace, they are spending big money to have access to the latest technology.

Microsoft and Coca-Cola joined together in 2020, but since then the capabilities of AI have grown exponentially — a big reason why Coca-Cola is committing to spend nearly five times more than it did the first time.

"This new agreement builds on the success of Coca-Cola's partnership strategy with Microsoft, showing our commitment to ongoing digital transformation," John Murphy, president and chief financial officer at Coca-Cola, said in a statement.

In 2023, Coca-Cola named Pratik Thakar as its global head of generative AI. The Sprite, Diet Coke and BodyArmor maker is using AI as a tool to connect with consumers. In 2023, it launched Coca-Cola Y3000, a beverage the company said is the first flavor co-created with humans and AI. It also has used the technology to launch several experiential marketing campaigns.

Coca-Cola is not the only company utilizing AI. Mars, AB InBev and McCormick & Co. also are incorporating it into their operations.

Dive Brief:

• GXO Logistics has partnered with robotics developer Dexterity to pilot AI-enhanced robotics to streamline warehouse operations for a beauty brand.
• The pilot combines robotics with artificial intelligence and machine learning to de-palletize, label and re-palletize packages, according to a February news release.
• ”Because these smart robots utilize AI, they require virtually no instructions or setup prior to implementation, and they self-train on the job, improving their performance with every pick,” the release said.

Dive Insight:

The critical need for a better level of warehouse efficiency in a “tougher environment” is what is driving the need for automation across nearly every vertical, Chief Strategy Officer Neil Shelton told Supply Chain Dive in an interview.

When volume growth is declining, as seen through the second half of 2023, “that’s quite often a catalyst for a company to turn around and look at its operations and the problems of supply chain challenges, wage inflation and supply chain inflation on the whole,” Shelton said.

The AI packaging pilot is an example “of where AI is driving a piece of automation technology to perform at a much higher ROI or level of performance,” said Shelton. “And it’s not just in packaging activities, we’re seeing an AI overlay for an awful lot of automation.”

While it is unclear what beauty brand GXO is piloting the Dexterity robotics for, the warehouse logistics company works with clients like Avon and Glossier.

The trial is part of the warehouse logistics company’s push to pilot solutions that boost productivity and workforce management.

GXO has been exploring AI applications for several of its clients, including a proof-of-concept pilot for Agility Robotics’ human-like robot “Digit,” which is currently being tested at Spanx’s facility in Georgia, according to a Dec. 6 press release.

Earlier in October, GXO inked an agreement with Eddie Bauer to optimize warehouse operations, which includes a fully automated 385,000-square-foot facility expected to be complete in the first half of this year, per a press release. The automated site will deploy autobagger packing machines and a totes-to-person picking solution using GreyOrange technology.

In July, GXO piloted a 3D sortation system where more than a dozen 3D sortation robots were programmed to feed 30 trolleys with more than 450 destination bins, according to a press release.

GXO plans for more of the AI enabled applications going forward.

“It’s not so much any one specific piece of technology, but rather the integration of numerous technologies to create an end-to-end product through the sites.”