AI adoption in supply chains has not yet reached its full potential. Santova’s Arne Walter talks about how collaborative efforts and organisational changes are essential for unlocking AI’s transformative potential.
It’s hard to read any news without encountering discussions about Artificial Intelligence (AI), especially in supply chain management.
While there’s a perception that AI is omnipresent in supply chains, the reality differs, as highlighted by recent industry surveys. According to a Deloitte and MHI survey, 85 per cent of respondents believe AI will reshape supply chains in the coming years, yet only 27 per cent are currently using AI tools, revealing a substantial 58-point adoption gap.
While adoption levels have increased over recent years, the gap has widened from a 48-point base in 2020 (MHI 2020 report: 60 per cent expected AI to reshape supply chains in the coming years/12 per cent actual usage). It seems AI adoption levels in the industry cannot keep up with the expectations of the technology by managers and supply chain leaders.
Why is it then that there is such a large gap between perceived utility and the implementation of AI solutions in supply chains to date?
“We have been involved in research on AI adoption in automotive supply chains in Germany, in cooperation with the Royal Melbourne Institute of Technology (RMIT) with Professor Kamrul Ahsan and Professor Shams Rahman”, says Arne Walter, Director, Santova Logistics.
“Part of this research tries to answer exactly this question by asking how companies’ structure organisational and technology resources should adopt AI. The study has analysed 11 automotive suppliers to investigate how these companies have implemented AI projects in their supply chains. The results of the study also offer interesting insights into the barriers to AI adoption that persist in the industry.”
These barriers help to explain the AI adoption gap that we continue to observe in surveys of supply chain leaders resulting in supply chains being unable to fully leverage AI technology and its associated productivity gains, according to Arne.
“We need to start by fencing in the terminology, as part of the problem certainly lies in what we refer to as AI,” he said. “This is partly also due to the recent hype around the technology.”
AI applications can be distinguished into four general groups: descriptive, predictive, prescriptive, and cognitive.
Descriptive AI applications include diagnostics; for example image recognition to spot quality issues in manufacturing. Predictive AI makes use of large datasets to learn and make predictions for the future. Typical applications for predictive AI can be found in supply chain demand planning and predicting future demand.
Prescriptive AI goes one step further by analysing large datasets to find an optimum between different scenarios. These applications can be used to plan advertising spent between different media platforms to maximise the impact of dollars spent or plan product-store allocations to maximise revenue by placing products in stores or within the store to maximise sales. These applications often also include demand planning at the distribution centre or store level.
Predictive and prescriptive AI are often implemented in conjunction with other technologies in supply chains, like advanced robotics in automated warehousing for example.
Cognitive AI is best known as generative AI, machines that seem truly intelligent. Most cognitive AI applications utilise the same technology as predictive AI as they are based on pattern recognition also used by the large language models that power generative AI. Generative AI applications are ubiquitous but using them in supply chain management is still limited.
What inhibits organisations from implementing AI tools? The reasons could be varied. Recent research done by Santova along with RMIT identifies some factors.
Technology often seems somewhat experimental making it difficult to establish a return on investment at the outset of the project. Despite all the hype, once it comes down to putting money on AI, many company boards hesitate as they fail to quantify their value. Use cases or small pilots can help to demonstrate the value of AI technology on a small scale.
Hand-in-hand with these issues goes management’s understanding of the technology. Often driven by technology hype, decision-makers are unclear about the tangible benefits of AI in the context of their company.
A lack of affordable AI talent and data availability in the right context, quality and quantity.
These points can be summarised in the key findings seen in this research, confirming findings from other studies. Technology adoption like AI is not a technology issue. It is foremost an organisational issue and requires organisational change. This includes leadership understanding of the technology to make the appropriate decisions and steer the business through the technology adoption process.
All these issues compound in an ongoing lack of AI adoption in supply chains, which seems out of sync with the sky-high expectations and possibilities the technology offers.
What can be done to overcome the adoption gap?
Many companies are turning to external parties to support them. Specialised companies can provide AI expertise and access to cloud-based computing power that is required to run AI models. Santova Logistics, as a technology-based international trade solutions provider, is doing exactly this for its clients.
Santova is empowering its clients by providing them with access to AI technology that allows them to manage their supply chains smarter and more cost-efficiently without the need to invest in AI technology directly.
Its predictive ETA AI tool uses deep learning technology to provide real-time predictions of vessel ETAs. On a transactional level, this allows clients to anticipate ETA changes in advance and implement mitigation strategies that result in real cost savings.
On a strategic level, Santova provides predictive analytics tools to customers that allow them to create a digital twin of their supply chain that can be analysed to identify efficiencies on various levels.
The tools are instrumental in exposing inefficiencies and opportunities for cost savings. The differentiator is in the implementation – it is relatively easy to analyse a dataset and identify improvement opportunities within a client’s supply chain. This is where most advisory service providers stop.
Santova adds a layer of understanding, working with the customer to implement change. Santova recently identified considerable potential cost savings by consolidating LCL cargo from various suppliers using predictive data analytics for a large client – that was the easy part.
Implementation of these initiatives required a close working relationship with the client, suppliers and service providers through the process of Inco term changes, procurement negotiations and workflow setup resulting in quantifiable cost and lead time savings.
The gap between AI’s potential use and current adoption in supply chains persists. Challenges like AI’s perceived experimental nature, ROI ambiguity, and leadership understanding hinder widespread implementation. Santova Logistics follows a solution-oriented approach, offering AI tools for real-time decision-making and cost savings without extensive internal investments by their clients.
For more information on Santova, click here.