The Sourcing Elite Board (SEB), established by Global Sources, is an elite membership club limited to invited top industry professionals. Currently, it has over 50 members, mainly senior managers in sourcing from Shanghai and Hong Kong, as well as professionals with academic backgrounds. SEB holds periodic sharing events within the club, inviting members to discuss sourcing strategies, e-commerce innovations, and insights and forecasts on the global economic landscape. It aims to inspire innovation and drive industry advancements, providing a vibrant communication platform for professionals and experts in sourcing.
Recently, Chief Executive China invited Erik Walenza-Slabe from the Global Sources SEB community for an interview. Erik shared his insights on the disruptive impact and risks of applying AI technology in enterprises, particularly in supply chain companies.
Erik Walenza-Slabe is the CEO of Asia Growth Partners, the Managing Director of the digital consulting firm IOT One, and an active member of the Industrial Internet Consortium. He also leads the Technology Innovation Committee of the American Chamber of Commerce. Since 2013, he has managed the Startup Grind Shanghai chapter and founded the IOT One Academy in 2015 to help companies overcome capability gaps in their digital transformation processes.
Chief Executive China (CEC):What do you think is the most disruptive impact of AI technology on supply chain management?
Erik Walenza-Slabe (EW):I believe the most disruptive impact of AI on supply chain management is its ability to process unstructured data. Traditional IT systems excel at handling structured data, such as information from ERP, CRM, or other supply chain management systems. However, most relevant information in the world is unstructured, such as news articles, press releases from suppliers or customers, and emails or messages with partners. Currently, this information can only be processed manually.
Of course, labor costs are high, and time is limited, so much information goes unprocessed or is addressed too late. We might learn about something a week later, making timely responses impossible. Now, AI can transform unstructured data into structured data for decision-making. For instance, if there’s a press release about a product recall, we don’t have to wait for team members to happen upon it; we can scan relevant information daily. Once the system captures such information, it can immediately identify recall events and respond quickly.
This will fundamentally change our ability to respond to incidents and optimize decisions. This change is still in its early stages, but many traditional supply chain management software companies have begun integrating unstructured data. Furthermore, many native AI software companies focused on this issue are emerging.
CEC: So, will this make us more efficient?
EW: Yes, efficiency improvement is one of the impacts.
For example, a few years ago, we worked on a project for a chemical company with a large team but low accuracy in forecasting. We tried to identify all possible variables affecting the supply and demand of their chemicals, which could significantly impact market prices, as these prices might fluctuate dramatically weekly. Many critical variables involved changes in consumer sentiment or willingness to buy cars or homes in the U.S., initially just sporadic data points. However, it was challenging to track such unstructured data at the time.
Now, AI can analyze these trends (like demand in the U.S. real estate market) efficiently and measure potential impacts on polycarbonate prices (related to the chemical company), thus optimizing forecasting and pricing decisions.
Additionally, AI can enhance business insights. As I mentioned earlier about competitor product recalls, you must be able to capture that information in a timely manner to call customers the next day and pitch alternatives. This ability to extract actionable information from unstructured data is transformative.
CEC: What is the most underrated improvement AI technology can bring to enterprises, especially supply chain companies?
EW: In the past, technological advancements were typically led by industries and enterprises, with consumer markets following suit. Now, the situation has reversed. The consumer market is leading, while industries have become followers. This is particularly true in the supply chain field, where the internal complexities and risk awareness often mean technologies are adopted only after they mature.
In the next five years, as companies gradually address internal organizational challenges (like cybersecurity, privacy, and process restructuring), AI’s capabilities will rapidly advance.
In the past two years, we have witnessed explosive growth in new AI capabilities, especially in decision-making. AI can now reach the level of human managers in most decision-making scenarios. For instance, AI can provide more accurate diagnoses than doctors based on patient symptom descriptions.
In supply chains, the main obstacles are organizational rather than technical. Once companies resolve these issues, the impact of AI may exceed their expectations, allowing for faster processing of vast amounts of information, optimizing supply distribution, vendor selection, and pricing decisions. It might even replace mid-level management decisions with AI proposals subject to human approval.
CEC: But won’t this be a daunting scenario? If machines can decide for themselves and produce independently, why would companies still need human employees?
EW: I acknowledge that this is indeed unsettling. From a legal perspective, we have already seen this trend, as the legal field is, in some ways, simpler than supply chains. Supply chains require the integration of many different systems and involve complex interpersonal relationships, while much of the work in law firms hinges on understanding legal systems and historical cases, identifying relevant clauses from client descriptions. This process is relatively straightforward.
Now, we find that law firms are significantly reducing the hiring of junior staff because AI can replace basic legal documentation work. However, senior lawyers remain indispensable since they manage interpersonal relationships. Thus, the team structure in law firms may shift from "10 senior + 90 junior" to "15 senior + 15 junior."
In fact, a significant risk is that humans may lose control over the decision-making process. For human errors in decision-making, we can at least find responsible individuals to communicate with and hear their reasoning. But AI may make decisions based on thousands of variables that humans find difficult to understand. Therefore, we need to find a balance between improving decision-making efficiency and maintaining human control over the system.
CEC: What is the most common capability gap when enterprises apply AI technology? What might be the main reasons for project failures?
EW: Small and startup companies can quickly adopt AI, but large enterprises face multiple challenges: outdated systems are difficult to revamp, internal governance rules are rigid, employees resist AI due to fears of layoffs, and traditional companies struggle to attract high-salary AI experts.
For instance, a large European flavor company initially planned to recruit high-paid talent from tech giants to form its digital team, but AI talent generally prefers not to have traditional manufacturing companies on their resumes. They had to offer a 30% pay increase to attract candidates, but this proposal was rejected by their HR department due to salary rules. Ultimately, they had to outsource AI development, meaning that core advantages like capability building and data accumulation are held by external companies.
Additionally, the cognitive gap at the decision-making level is a key barrier: most decision-makers in traditional large enterprises are over 50 years old, and their careers were formed in different technological eras, often struggling to accurately assess the maturity of AI technology. For multinational companies operating in China, there is an added geographical gap—feedback regarding market demands and technological developments in China needs to reach the European headquarters. This cross-border communication is compounded by considerations of cybersecurity, privacy process changes, and significant investment decisions, making AI deployment decisions for foreign enterprises in China even more challenging.
Moreover, many companies rely on "approved but inefficient" tools like Microsoft Copilot, rather than vertical solutions. Thus, many lower-cost, targeted high-quality solutions from vendors are often not adopted because they are not on the pre-approved list.
Companies must confront these challenges and carefully plan their AI implementation paths.
CEC: What common traits do successful companies that have effectively applied AI technology share?
EW: Based on my observations of successful cases in the Chinese market, particularly in service to multinational companies, effective applications of AI typically have the following characteristics:
- First, the goals set can be ambitious, but the boundaries must be clear. Focus should be on transformative impacts on specific functions or processes, thus clarifying stakeholders.
- Second, establish empowered teams supported by executives to ensure high autonomy.
- Third, gain grassroots support. Engage frontline employees in communication about needs and co-design process transformations. Since frontline employees understand the actual business operations best, they can effectively help define functional requirements.
This combination of "focused areas + ambitious goals + top-level empowerment + collaboration" is significantly effective.
Conversely, a common failure pattern is vague discussions about "company-wide AI," lacking specific implementation plans, unclear responsibilities, ambiguous budget sources, and missing decision-making mechanisms. Many companies are satisfied with creating the illusion of AI applications without genuinely transforming business operations, often settling for solutions like Microsoft Copilot that are "compliant and easy to use but limited in effectiveness."
CEC: So, is the key to success relying on people rather than solely on AI technology?
EW: Absolutely correct. At least for the foreseeable future, while AI can efficiently automate specific process steps, humans must decide which steps can be automated and how to adjust processes around that automation. Humans need to establish priorities, weigh pros and cons, and plan initial investment directions. Therefore, the role of humans remains crucial—I believe this situation will not change in the short term.
CEC: Will there be significant differences in how U.S. and Chinese enterprises apply AI technology for innovation? For example, in mindset?
EW: We need to consider separately the situation of U.S. companies in the U.S., U.S. companies in China, and Chinese companies in China.
U.S. companies have a mature AI ecosystem with a wide range of suppliers to choose from. While talent costs are high, talent resources are relatively abundant. Moreover, due to significant “peer pressure” among many companies, AI applications are quite rapid and proactive, with many companies making significant progress.
Chinese companies, although starting slightly later (about six months behind), have a tradition of quickly adopting new technologies—once they see the profit impact, their actions can be swift.
As for U.S. companies in China, or more broadly, foreign companies in China, they face unique challenges: various AI solutions developed by headquarters often do not fit well in China. While companies can develop parallel solutions specifically for their China branches, if the Chinese market contributes too little to overall revenue, it’s hard to justify the investment in AI development resources. Therefore, foreign companies’ AI applications need to focus on areas that can make a significant impact—because in overall AI application coverage, they struggle to compete with local Chinese companies and cannot match globally-led solutions from headquarters.
CEC: Can it be said that Chinese companies are more willing to pay for trial and error?
EW: Yes. Chinese companies place a higher emphasis on speed and results and have a higher tolerance for trial and error—as long as they are willing to bear the costs of mistakes, this trait is advantageous for AI applications.
CEC: What advantages do you see in applying AI technology to solve long-tail scenarios?
EW: Many long-tail scenarios are invisible to humans.
On one hand, human labor is costly; on the other hand, humans can only process a limited amount of information. Therefore, when humans handle such scenarios, they typically only focus on core variables and track them continuously. A person might be well aware of 10 key variables but oblivious to 100 others that could also influence the market—processing the remaining variables is too complex to analyze and communicate.
AI, however, can track thousands of variables simultaneously and identify potential impacts. For example, low-frequency events like a competitor's factory maintenance may be overlooked by humans, but AI can analyze data and trigger actions. We can train AI to handle vast amounts of information that humans find difficult to perceive, leading to valuable insights. For example, based on the operational dynamics of a competitor's factory, AI can relay targeted suggestions to sales, production, forecasting, and pricing teams.
However, it must be reiterated that these scenarios fall under the long tail category, requiring systems capable of large-scale tracking and processing of relevant data. This is precisely where artificial intelligence will play a significant role.
Established in 2022 by Global Sources, the Sourcing Elite Board (SEB) is a collaborative initiative dedicated to advancing the sourcing industry through shared expertise and innovative strategies. Senior executives, from buying offices to retailers and brands, are welcome to join this distinguished community.
