1
The Self-thinking Supply Chain
Abstract
Purpose: An emerging theme in the practitioner literature suggests that the supply chain of the future –
enabled especially by developments in ICT – will be autonomous and have predictive capabilities,
bringing significant efficiency gains in an increasingly complex and uncertain environment. This paper
endeavours to both (i) bridge the gap between the practitioner and academic literature on these topics and
(ii) contribute to both practice and theory by seeking to understand how such developments will help to
address key supply chain challenges and opportunities.
Design/methodology/approach: A multi-disciplinary, systematic literature review was conducted on
relevant concepts and capabilities. A total of 126 articles were reviewed covering the time period 1950-
2018.
Findings: Results show that both IoT and AI are the technologies most frequently associated with the
anticipated autonomous and predictive capabilities of future supply chains. In addition, the review
highlights a lacuna in how such technologies and capabilities help address key supply chain challenges
and opportunities. A new supply chain model is thus proposed, one with autonomous and predictive
capabilities: the self-thinking supply chain.
Originality/value: It is our hope that this novel concept, presented here for the first time in the academic
literature, will help both practitioners to craft appropriate future-proofed supply chain strategies and
provide the research community with a model (built upon multidisciplinary insights) for elucidating the
application of new digital technologies in the supply chain of the future. The self-thinking supply chain
has the potential in particular to help address some of today’s key supply chain challenges and
opportunities.
Keywords: self-thinking supply chain, Internet of Things, artificial intelligence, autonomous, predictive.
Paper type: Research paper.
1. Introduction
A large number of academic and practitioner publications have acknowledged that supply chain
management is undergoing significant changes due to the adoption of new digital technologies
(Capgemini, 2016; DHL, 2016; Wu et al., 2016; Haddud et al., 2017). Breakthroughs in several fields,
such as the Internet of Things (IoT), artificial intelligence (AI), robotics, autonomous vehicles, and
additive manufacturing are transforming all the steps in supply chain management (WEF, 2017). This is
taking place in the context of the Fourth Industrial Revolution, a revolution that is characterized by an
unprecedented advance in digital technology, and which is blurring the lines between the physical, digital,
and biological spheres (Schwab, 2016). Among the breakthroughs that characterize the Fourth Industrial
Revolution is the ability to collect and analyse massive amounts of data in an automated way, then use
this data for decision making and implement decisions in real time. Practitioner research suggests that
there will be more than 50 billion devices connected to the Internet by 2020 (Cisco, 2011), a trillion
sensors connected to and transmitting information to analytical platforms in the cloud, and 44 trillion
gigabytes generated (DHL, 2015). In this context, information that was previously created by people will
2
increasingly be machine-generated, while the entire supply chain will be connected, including parts,
products, and other smart objects used to monitor the supply chain (IBM, 2015). Based on these data,
supply chains will be able to make decisions more accurately and in real time, to optimize operations,
handle incidents that require risk-mitigation actions, avoid disruptions, and satisfy an increasingly volatile
demand (Calatayud, 2017).
Notably, many commentators argue that the supply chain of the future will be autonomous and have
predictive capabilities (IBM, 2015; DHL, 2016; WEF, 2017). This, they say, will bring significant
performance improvement in an increasingly complex and uncertain environment for supply chain
management. Indeed, supply chains currently face a variety of risks due to growing internationalization
and firm interconnection, higher demand volatility, and faster supply chain speed (Christopher and
Holweg, 2011 and 2017). Driven by new digital technologies, the supply chain of the future will
increasingly be self-aware, think by itself and require minimum, if any, human intervention to manage
risks. The self-thinking supply chain will continuously monitor supply chain performance by analysing
quintillion bytes of data generated by objects; forecast and identify risks; and automatically take actions
to prevent risks before they materialize. The supply chain will autonomously learn from these activities
and use such knowledge in future decisions. Importantly, large amounts of data and the use of powerful
analytical and simulation models will allow the supply chain to predict the future with minimum error and
take actions to, for example, address constant shifts in demand. The self-thinking supply chain will thus
push supply chain flexibility and agility to limits yet to be discovered (Calatayud, 2017).
Despite these promising benefits for supply chain management (SCM), literature on self-thinking supply
chain is scarce. The term is mentioned only infrequently in the practitioner literature in an attempt to
predict future SCM trends with the simultaneous adoption of different new digital technologies (DHL,
2016; Calatayud, 2017; IBM, 2017). In the academic literature, however, current research mainly focuses
on identifying the impact of a single new digital technology - such as IoT - on supply chain performance.
Therefore, in this paper we seek to understand, from both practical and theoretical perspectives, how
multiple digital technologies will shape future supply chains. The literature on automated, predictive and
self-thinking supply chains, and related concepts and capabilities, is reviewed. The insights from that
review are then considered in the context of the current understanding of supply chain strategy and a new
supply chain model – the self-thinking supply chain – is posited. The systematic literature review spans
disciplines such as Supply Chain Management, Computer Science, Engineering, and Economics. It is our
hope that this novel concept, described here for the first time in the academic literature, will help both
practitioners to craft appropriate future-proofed supply chain strategies and provide the research
community with a model (built upon multidisciplinary insights) for elucidating the application of new
digital technologies in the supply chain of the future.
This paper is organized as follows: Section 2 lays out the methodology and procedures followed to
conduct the systematic literature review; Section 3 presents the results of the systematic literature review;
Section 4 discusses the impact of new digital technologies on SCM according to the extant literature;
Section 5 introduces the self-thinking supply chain model and elucidates its contribution to supply chain
strategy; and Section 6 presents the conclusions and outlines areas for future research.
2. Methodology
In order to explore the characteristics of a self-thinking supply chain, the systematic literature review
technique was applied. This technique uses systematic methods to identify, select and critically evaluate
the body of knowledge related to a given topic (Gligor and Holcomb, 2012; Rousseau et al., 2008;
Tranfield et al., 2003). Unlike a traditional literature review, which might be influenced by the familiarity
or preferences of the reviewer, a systematic literature review allows the researcher to gather, analyse and
3
interpret a comprehensive body of available literature in a thorough and unbiased manner (Wang and
Notteboom, 2014).
The systematic review technique is particularly relevant to the purpose of this paper. By avoiding the
biases of conventional literature reviews, a systematic review allows the researcher to: (1) summarize the
accumulated body of knowledge related to the topic of interest; (2) explore the topic through different
perspectives; and (3) develop reliable knowledge from a pool of knowledge dispersed across a broad
range of studies (Gligor and Holcomb, 2012; Tranfield et al., 2003). Given that the pool of knowledge on
new digital technologies and supply chain management is spread across a variety of academic disciplines
and that, according to practitioner literature, a self-thinking supply chain encompasses the use of different
technologies, the systematic literature review is deemed appropriate to explore how multiple digital
technologies will shape future supply chains. Indeed, a systematic review of automated, predictive and
self-thinking supply chains, and related concepts and capabilities allows us to explore available academic
literature comprehensively, giving insights on the meaning, enablers and potential benefits of a self-
thinking supply chain, while bridging the gaps among different perspectives and developing a broad
understanding of the research topic.
Applying the systematic review technique involves five stages (Figure 1): (1) problem formulation; (2)
literature research; (3) selection and evaluation of literature; (4) research analysis and interpretation; and
(5) presentation of results (Denyer and Tranfield, 2009; Gligor and Holcomb, 2012). The problem
addressed in this paper was formulated as follows: given that the pool of knowledge on new digital
technologies and supply chain management is spread across a variety of academic disciplines, can we aim
to develop an integrated framework to understand the defining aspects of a self-thinking supply chain and
its potential benefits from both practical and theoretical perspectives. The literature was researched by
interrogating the dataset Scopus, one of the largest repositories of academic articles. Literature research
comprised five stages. In the first stage, keyword search was performed using the words (“self-thinking”)
AND (“supply chain”), together with related words such as (“autonomous” OR “predictive”) AND
(“supply chain”), in papers and conference proceedings published between 1950 — the earliest available
year in the dataset — and February 2018. In the second stage, studies were chosen and evaluated
according to a set of specific criteria that referred to: (1) the relevance of the study to the research
problem; and (2) the quality of the study. In agreement with Wang and Notteboom (2014), the Critical
Appraisal Skills Program (CASP) checklist was used to evaluate the quality of the studies. Studies
selected in stage two were analysed in order to identify shared patterns among them. The analysis showed
that studies could be grouped into two domains: (1) studies exploring the use of Internet of Things (IoT)
in SCM; and (2) studies exploring the use of artificial intelligence (AI) in SCM. In the third stage, the
dataset was further interrogated using keywords that referred to such domains. In the fourth stage, search
results were evaluated according to the relevance and quality criteria applied in stage two. References
included in the papers collected were used as guidance for further exploration of the literature. In
addition, literature citing the papers collected were identified and analysed. In all queries, words closely
related to self-thinking such as ‘smart’ or ‘intelligent’ were considered as well. In the fifth stage, the
review of articles was complemented by searching: (1) the catalogue of the United States Library of
Congress (the biggest library catalogue in the world) for books that could be related to the topic; and (2)
Google search engine, using the same keywords that were used in the Scopus query, to account for
working papers and reports relevant to the topic published by other sources, such as national and
international organizations. Search results were evaluated according to the relevance and quality criteria
applied in stage two.
4
3. Results
The first stage of the literature research resulted in 89 articles. In stage two, the 89 articles were evaluated
according to the relevance and quality criteria, with 28 articles satisfying such criteria (Table 1). Next, the
articles selected were preliminarily analysed with the objective of identifying shared characteristics that
could be used to group and classify them into different categories. The analysis showed that articles could
be classified into two broad domains: (1) articles exploring the use of IoT in SCM, including studies with
a focus on planning and management of activities that integrate supply and demand within and across
companies; (2) articles exploring the use of AI in SCM, including studies that develop and apply different
types of algorithms to dynamically solve supply chain optimization problems. These categories were then
used to further query the database, looking for articles relevant to the research problem.
[Table 1]
In the third stage, the database was interrogated by searching for words related to the two domains
identified in the previous phase: IoT and AI. The words (“Internet of Things” OR “IoT”) AND (“supply
chain”) were searched for the first domain, resulting in 397 articles, among which 56 articles satisfied the
relevance and quality criteria. Next, the words (“artificial intelligence” OR “machine learning”) AND
(“supply chain”) were selected for the second domain, resulting in 141 articles, among which 23 articles
satisfied the selected criteria. In addition, a third search was performed using the keywords (“Internet of
Things” OR “IoT”) AND (“artificial intelligence” OR “machine learning”) AND (“supply chain”) to
identify articles encompassing both types of technology, thus combining both domains. This search
resulted in 22 articles, among which 17 articles satisfied the selected criteria. Overall, after applying the
relevance and quality criteria to the results, 68 articles were selected, making up the basis for further
analysis. The earliest article included in the dataset had been published in 2007 and the most recent in
2018. This time period is consistent with the exponential growth of academic interest in the subject of
digital technologies applied to SCM. Indeed, the simple search for the keywords (“digital”) AND
(“supply chain”) on Scopus showed that 80% of the academic publications found (1,128 articles) were
published in the period 2007–2018. In addition to the articles selected for analysis, both references
contained in and literature citing these articles were analysed and also included when they satisfied the
selected criteria.
In the fifth stage, the literature research was complemented by querying the catalogue of the United States
Library of Congress and Google search engine, using the same keywords from previous phases. Arising
from the five stages of the literature research, 126 studies were selected and analysed (Figure 1).
5
Figure 1. Graphical illustration of the literature search process and results
Table 2 shows the five journals with the highest number of articles selected through the literature search
process.
[Table 2]
4. IoT, Artificial Intelligence and SCM
SCM aims to get, in the right way, the right product, in the right quantity and right quality, in the right
place at the right time, for the right customer at the right cost (Mangan and Lalwani, 2016). However,
growing supply chain complexity, higher demand volatility, unprecedented technological changes, and
supply chain speed are making SCM increasingly challenging (Christopher and Holweg, 2017; Fore et al.,
2017). In 2017 32% of S&P 500 companies were affected by supply chain disruptions (Resilinc, 2018).
To overcome supply chain risks and vulnerabilities, academic and practitioner literature suggests that
