How will GenAI prompt a step change toward autonomous supply chains?

Recognizing its potential, companies see GenAI as a critical capability for remaining competitive in the future and are investing in it accordingly. Three-quarters are planning to deploy GenAI in their supply chains and 80% believe it can reinvent supply chains and are making it a high priority. In addition, 69% believe that failing to integrate GenAI into their supply chains will put them at a competitive disadvantage. 

Today’s use cases

Most organizations’ supply chains (82%) are using both AI and GenAI across a wide array of use cases to take advantage of their different, often complementary strengths. Traditional AI is rules-based, requiring prepared data sets and predefined logic to solve business problems. GenAI is great for text-rich environments and unstructured data, creating new content based on the data it has been trained on.

For example, companies using traditional AI for demand forecasting and quality optimization are finding that a GenAI layer improves accuracy and democratizes adoption of the tools.

“The implementation of GenAI has improved the accuracy of demand forecasting by 50% in the last six months,” says the Head of Distribution at a global consumer electronics company. “We’re seeing considerable success across the supply chain.”

These are areas where traditional AI has long been available but has been limited by the need for highly trained data scientists to stand up and maintain, placing them out of reach and too expensive for many. The benefit of GenAI is in providing a natural language, interpretive layer that can become a democratizing force that puts these tools in the hands of the workforce. These areas are also supply chain functions with well-defined datasets, a high percentage of unstructured data and high value to be gained.

The next wave of use cases, those with lower GenAI deployment today but high anticipated use in two years, potentially includes:

• Supplier management (purchasing)
• Production yield or quality optimization (manufacturing)
• Risk management (planning)
• Customer service chatbots and product training (pre/post sales)

For example, customer service chatbots and product training are language-rich environments where organizations can train GenAI on voice recordings and transcripts from previous customer calls to rapidly expand the number of queries that chatbots can resolve. With product training, GenAI can massively speed access to key information that is stored in product manuals.

These areas of high growth also offer clear commercial returns, either by improving the speed of customer service through chatbots or reducing costs through quality optimization in manufacturing. Demand forecasting is another key area of focus for GenAI. It’s perceived to be a use case that will resolve a lot of pain points for the supply chain and offers clear metrics that can make the business case an easier sell to the CEO and the Board.  

Many of these use cases suffer from back-end challenges when using traditional AI (quality optimization, predictive maintenance) and require bespoke solutions. Customer service chatbots also have a high degree of risk given the chatbots’ direct interaction with the public. Both factors may be contributing to the time horizon suggested here.

When comparing anticipated growth in GenAI in two years, data from the survey shows that Front-runners are more likely to anticipate quicker progression of GenAI implementation in logistics, manufacturing and pre/post sales when compared with Followers. This is likely because the Front-runners are already seeing more success in other areas, are more integrated end-to-end and so have a better data foundation, leading to more confidence to go further faster and accelerating the gap from Followers. For example, logistics network design is an area that has evolved from a one-off project every five years to a dynamic task that is reassessed every six months. Leading companies are building this capability to future proof their organizations.

Examples of GenAI use today

Implementing GenAI in the supply chain involves a complex interplay of technical, organizational and operational challenges. Organizations should take the following action steps to advance their journey to the autonomous supply chain and exploit the full potential of GenAI.

1. Align people and investments to strategic vision

For Front-runners, the top factors for success in GenAI deployments are securing support from leadership (67%), building support from third parties (65%) and availability of technical talent (64%). The largest gaps between Front-runners and Followers are in prioritizing the strategic vision (62% of Front-runners versus 39% of Followers) and support from third parties (65% of Front-runners versus 47% of Followers), which highlight the importance of a cohesive vision and external support in ensuring GenAI pilots and implementations are focused on delivering business value.

“My advice is to do your homework. Is the newest thing on the block really going to add value?” says the Global Operations Director of a consumer products manufacturer. “And vet external partners thoroughly on a technical level to verify their claims.”

Skills gaps are a recognized challenge during implementation for survey respondents, coming second after data readiness. Human factors will dictate the speed of adoption, so robust plans for upskilling, alongside targeted talent and hiring plans, are essential. Front-runners are tackling this by putting GenAI with appropriate controls into the hands of their employees (a priority for 51% of Front-runners versus 33% of Followers). This is especially pertinent given the findings from the EY Work Reimagined Survey, where nearly half (49%) of employees are already using, or expect to use, GenAI in the next 12 months. However, few (17% of employees and 22% of employers) are making training in GenAI skills a top priority.

A cohesive, strategic vision can clarify investment priorities against the seemingly endless list of use cases of GenAI, minimize the risk of multiple business units duplicating investments and improve AI outcomes by guiding the augmentation of large language models (LLMs) with reusable algorithmic patterns, such as retrieval-augmented generation. 

2. Prioritize data readiness when considering use cases

The demands of GenAI are shining a spotlight on the myriad complexities of enterprise data management. Despite data availability, quality and privacy being top of mind when prioritizing use cases, organizations are still struggling. Maintaining data quality is the number one implementation challenge cited by respondents (38%), with access to data (33%) a top challenge as well.

This highlights the data challenges particular to supply chains, where data is fragmented across different systems in the organization and across external parties. Resulting issues include massive volumes, incomplete data, data privacy regulations, data architecture, integration with other technologies, access management and other interconnected risks.

Given the fundamental importance of data to AI, it is no surprise that Front-runners are data champions.

“We see GenAI as key to navigating the huge volumes of customer data we possess and mining it for insights,” says the Head of Distribution at a global consumer electronics company. “GenAI revealed that we were storing half a billion records on customers who hadn’t interacted with the company beyond the initial sale.”

Putting in place the right data architecture, tools and policies is an investment worth making, because these capabilities are foundational to the effective deployment of GenAI.

“You need real-time datasets for prompting predictive algorithms,” says the CIO of a global electronics manufacturer. “But we gather data monthly; we don’t have a real-time data tempo in the supply chain yet.”

Any organization hoping to compete in GenAI needs to get its data house in order, by prioritizing data cleansing, standardization, systems and engineering to reduce latency, and enhancing metadata so data can be consumed by Retrieval Augmented Generation (RAG) systems to improve the accuracy of GenAI outcomes.

3. Maximize GenAI value by mitigating cyber and data risks

GenAI is a nascent technology, so it is not surprising to find 40% of respondents saying their organizations do not fully understand the new risks and challenges of GenAI in supply chain. While this should rise over time, there are two sides to maximizing value from GenAI initiatives: reducing risks and optimizing opportunities.

GenAI poses new vulnerabilities, for example through prompt injection attacks designed to provoke LLMs into leaking sensitive data or manipulate their outcomes. Front-runners are more likely than Followers to focus on the new risks that GenAI poses, such as inaccuracies and hallucinations, exposure to legal liability through IP infringement, overreliance on untested technology, brand or reputation damage and job insecurity.

“Our GenAI chatbot pilot performed well, but hallucinations crept in after a model update,” says Head of Technology of a global online marketplace. “We reassessed each use case, changed models for some and discarded others that no longer added value.”

The need for stronger cybersecurity, a top risk and implementation challenge in its own right, is paramount as organizations look to deploy GenAI. Topic clustering analysis from the 2024 Global Cybersecurity Leadership Insights Study shows supply chain vulnerabilities have doubled over the past five years. While cybersecurity functions are expanding their use of AI to stay ahead of threats, supply chain and operations leaders need to work closely with cybersecurity teams from the beginning to help with the secure adoption of GenAI in the supply chain. This includes embedding the cyber team in use case identification and governance to ensure the value potential of GenAI in the supply chain is maximized. 

Finding your North Star

The dream of an autonomous supply chain runs the gamut of reactions from operations and supply chain leaders — from excitement about the step change that GenAI represents to skepticism that full autonomy will ever be possible end-to-end.

Still, it represents the pinnacle of achievement. Even if organizations never reach end-to-end autonomy, it is a worthwhile “North Star” goal to align with an organization’s strategy, because every step closer to autonomous supply chains boosts resiliency by improving visibility and agility. Automation is likely to continue as it has done to date, piece by piece, pocket by pocket.

GenAI is already proving it can be a game-changer for the supply chain, and this nascent technology is expected to establish itself firmly in the supply chain over the next two years. For organizations to realize the full potential of GenAI, agility will be key.

Summary

GenAI is emerging as a transformative tool, enabling more autonomous supply chains with end-to-end visibility and real-time problem-solving capabilities. EY research suggests that GenAI adoption is crucial for competitiveness, but challenges in understanding risks and implementation complexities have led to a strategic reassessment. Front-runners in GenAI adoption are leveraging it for improved demand forecasting and operational efficiency, while others lag behind. To overcome these challenges, companies must align strategic visions, prioritize data readiness and mitigate cyber risks.

Gautam Jaggi, AnnMarie Pino and West Coghlan of EY Insights, Ernst & Young LLP, and Joe Morecroft of EY Brand, Marketing and Communications, EYGS LLP, contributed to this article.