KITT4SME report 2021: Artificial Intelligence adoption in European Small Medium Enterprises


In the first months of the KITT4SME project, SUPSI, WUT and Ginkgo Analytics collaborate to realise the KITT4SME report 2021. This report includes: - a sum-up of a methodology to assess AI readiness and maturity level in SMEs; - the results obtained through a survey investigating AI adoption that involved 36 European manufacturing companies; - a set of guidelines supporting AI adoption in SMEs.
KITT4SME report 2021
Artificial Intelligence adoption in
European Small Medium Enterprises:
The enormous transformation brought by
the fourth industrial revolution in the world
of production has forced companies of
all sizes and sectors to embark on the
digitalisation journey. In this context, the
exploitation of data to solve problems of
disparate nature through Artificial
Intelligence (AI) systems is rapidly
catching on but this requires
economically and technologically
accessible solutions for mass use.
The results of our survey show that both
SMEs and large enterprises are struggling
with the introduction of AI. This is mainly
due to poor digitisation. In fact, nearly 1/3
of the companies state that data
collection is an issue that does not allow
them to exploit AI. AI readiness is also
influenced by the size of the company
and the sector. In general, large
enterprises are more prepared than SMEs
and, in some sectors, AI is quite
widespread compared to others.
The survey is part of the activities of the
KITT4SME project, a European-funded
project which specifically targets
European SMEs and mid-caps to provide
them with scope-tailored and industry-
ready hardware, software and
organisational kits, delivered as a
modularly customizable digital platform,
that seamlessly introduces AI in their
production systems.
For more info, please contact
Nearly 1/2
of companies do not
have any AI solutions
or applications in
of respondents are
defining a strategy to
promote AI adoption
within the company.
of companies believe
that the collection of
data is fundamental
for their business.
Only 1/10
of companies are
training their
employees on digital
Starting from November
2020, the KITT4SME
project surveyed 36
companies in the
manufacturing field on
AI adoption and
readiness. The
proposed survey did not
focus only on the
operational aspects but
also, at a higher level,
on how organization
and culture inside a
company allow to fully exploit the potential of AI. The model is structured in five
pillars, each representing a company area that needs to be developed for the
adoption of AI. These pillars are Digital and Smart Factory, Data Strategy, Human
Resources, Organisational Structure, Organisation’s Culture.
Digital and Smart Factory - Thanks to digitisation of the real world,
information can be displayed, analysed, stored and used to make
better decisions that generate benefits for operations and
business. When AI is used to support or even is in charge of
decision making, it brings its maximum potential and value to
Data Strategy - Without data, AI cannot be implemented. AI
solutions take data from the factory environment as input and are
able to extract useful information, optimise operations and sug
gest or even make decisions. A proper Data Strategy has to guide
data management, data governance and proper infrastructure
Human Resources - As AI solutions spread through company
processes, the workforce has to master certain competencies in
order to make the most out of it. Companies must also help their
employees to acquire integrated, interdisciplinary IT skills that
provide them with a basic understanding of the applications and
processes used in different parts of the business, thereby
upgrading long-established occupational profiles.
Figure 1 AI maturity and adoption model
Organisational Structure - The adoption of AI inside a company
requires, like most of the ICT projects, the right organisational
structure. The transition to AI-based organisations involves a left-
to-right rethinking of how organisations run their operations,
engage their customers, empower their employees, and even
define their products.
Organisation’s Culture - Experience with lean management in the
1990s and 2000s showed that the key to its successful
implementation throughout a business is to change the mentality
of its employees. The same applies to the digital transformation
and AI adoption. Companies will be unable to achieve the
benefits if they simply introduce technologies without also
addressing their corporate culture.
Companies by size
Different industries
small enterprises
medium enterprises
large enterprises
Food and Beverage
Steel production and
AI technology is not widely adopted in
the manufacturing field and those
implementing it, use it only for single
applications, e.g. quality control.
However, only 10% state that they do
not want to use AI. More than 35%
desire to use AI along entire processes,
and 20% across the entire enterprise.
Digitisation is fundamental for a
company to tackle the path towards
complete automation of its processes
in the best possible way, involving AI as
the main promoter. However, the level
of digitisation of both SMEs and large
companies, are barely sufficient,
making AI implementation difficult.
Industry 4.0 is well known and many of
the respondents state that they heard
more than just the concept. This is
different for AI, with only 15% saying
they are familiar with it.
While 50% of companies do not have
any AI strategy at all, several have
started to define or even have already
in place pilot projects mainly in the
field of quality control, planning and
Although many of the businesses
define the data strategy as
fundamental, data collection is still
mainly done manually. Few
companies have automatic collection
systems avoiding duplication and
aggregation issues. Another problem is
that companies are also struggling to
collect data that they would like to
Only one company in the sample
declared it shares data all along the
supply chain, while the majority
manages operations internally without
sharing data with partners. However,
sharing data with suppliers and
customers is essential to understand
market trends and to discover possible
The lack of sharing is mainly due to:
data protection issues
absence of trust in other
not seeing a real value in doing
absence of an appropriate
According to the results of the survey and to several face-to-face interviews, five
key barriers prevent companies from adopting AI:
Lack of data: companies do not have enough data to feed AI. Many are still
collecting data manually or not collecting them at all. Since AI needs huge amounts
of data, reliable and continuous data collection and storage is fundamental to find
patterns and train AI systems. Without data, it is very difficult for algorithms to
discover the desired insights allowing the company to improve. Therefore,
structuring and automating data collection has to be a priority for companies willing
to adopt AI.
Lack of AI lifecycle assessment methods: companies have the perception that the
cost of AI solutions is still too high. This could be true in many cases, but it should be
compared with the achievable benefits. However, there is the lack of AI-specific
methods and tools to estimate the cost/advantage ratio and the payback time.
Moreover, companies struggle in defining a clear path to AI adoption making the
definition of investment risks even more complex.
Lack of customised solutions: companies struggle in finding personalised AI solutions
that can solve their problems at an affordable price. AI systems available on the
existing marketplaces, or those directly offered by software houses, may miss
features that companies need or, to the contrary, offer too many that increase the
price but remain unused.
Under-skilled employees: employees are not ready for the AI transformation. They
need more training for producing value for the company out of AI usage. Digital
skills are important and having a strategy for
measuring the current gap and for providing the
needed upskilling is an unescapable requirement.
Complexity of solutions: AI tools are still too
complicated. Companies need simple systems
that can be deployed and used
quickly by employees avoiding
complex setups, commands or
user interactions. An easy-to-use
system can avoid non-value-
added activities by adopting
simple operations to obtain a
clear and expected result.
The survey and the analysis of its results
allowed to identify 4 different
organisation archetypes.
Organisations that have not expressed
any interest in AI and have not yet
moved in any direction. Either AI is
unable to solve their problems or to
generate new opportunities, or, most
probably, they are truly lagging
behind everyone else. Either way, they
need to move! The fields of application
of AI are innumerable.
Organisations that have touched the
practical aspects of digitisation and AI,
and have already
exploited them. However,
they lack a high-level
overview, including
strategy and culture. When things get
complicated, their flaky foundations
may start shaking.
Organisations that have expressed
interest in AI and are already
implementing an effective data
strategy. However, there are still some
steps for reaching the most advanced
companies. A good idea
could be to learn from
them to make the final
step shortly.
Organisations that look well prepared
and have already adopted AI. They
have already a clear and defined
strategy, aiming at getting the most
from digital and AI solutions. However,
they should not rest, others may reach
their readiness level shortly!
Which archetype do you belong to?
Take part to our survey to receive your
custom assessment
Our research team has identified different actions that can guide companies in
embracing AI. These actions have been classified as:
A good start is half the work: actions to start walking in the AI and digitalisation
Competition stimulates the achievement of great results: actions to ride inside
the peloton of AI adopters.
AI is a competitive advantage: actions to boost and join the AI leaders to get
the most from this technology.
Plan your digitalisation journey
Plan and program a digitalisation journey to embrace the Industry 4.0 paradigm
within the organisation. As a first action, organise a workshop involving all the
internal stakeholders in order to identify the most significant challenges, existent
problems and current level of automation. Starting from these elements, prioritise
actions and activities, analyse existent use-cases and define a digitalisation
supervisor together with a team in charge of the operational activities.
Sneak a peek from the outside and seek for help
Involve external consultants and experts to put supposedly unchangeable
processes up for discussion and to bring in influences and ideas from already
realized projects from other companies. Analyse your industry competitors or
partners that have already introduced AI to get inspiration from their success stories.
Identify simple use cases and experiment by investigating from practical problems
and from which begin your journey towards AI adoption realising the first pilot
applications and experiments.
Diagnose your challenges
Diagnose the challenges and the problems that persist underneath the process
surface and that hinder the achievement of your desired performances. Start to
analyse most recurring non-conformities and customer complaints.
Involve your stakeholders
Interview the different stakeholders in the organisation and use structured methods
to analyse the most critical issues. Put employees on the front line and teach them
not to hide problems, but to face them.
Identify where the value is
Start from AI systems you have already developed or integrated to extend them to
all the most relevant processes. Apply AI only when it brings concrete value to your
processes. Assess the potential business value of the various possible applications
you have identified and choose the ones that are most suitable for you.
Widen your vision
Start to think possible AI applications also collaborating with your partners and
suppliers, considering the possibility to collect data not only from your production
process but also from the entire supply chain.
Extend starting from your success stories
Start from AI systems that you have already integrated in the production system to
extend them to all the most relevant processes.
Define your pipeline and proceed at full speed
Set up an AI use case pipeline (ideation, validation, prototyping, pilot, rollout).
Continuously iterate, improve use cases and build new ones.
Identify valuable data
Identify most relevant data that can provide useful information on processes and
foresee future use. Start specific use-cases for the prioritisation of data sources to
implement first, before implementing huge data lakes.
Do not build when you can re-use
Try to use as many existing tools as possible to gain speed (do not reinvent the
wheel). Today’s cloud computing platforms already have a lot of use cases
Ensure quality and quantity
Make the data collection structured, and when possible automatic, to avoid
duplicates, human error and non-value-added activities. Collect data using efficient
and effective methods to be ready to realise AI models that lead to precise and
accurate results.
Plan the data management
Develop a data management strategy that allows to clean and transform the raw
data into useful information which can be analysed and used to derive possible
insights. Think where to store all these data, on-premise and in-cloud solutions have
different pros and cons; choose the solution that most fits your needs.
Define your goals
Promote the collaboration between business and IT department in order to first
define business requirements and then services and infrastructure specifications.
Build your infrastructure
Deploy your ICT infrastructure and technologies (data platform, cloud services,
processing units, connectivity, bandwidth, etc.) in order to effectively collect,
manage, store and use data.
Protect data
Start to define a proper data governance to secure data from leakages, attacks
and even unwanted access by unauthorized people within your organisation.
Define data ownership in order to account for the quality of your data sets.
Define the data infrastructure
Define a proper data architecture collecting models, policies or standards that
defines which data are collected, how it is stored, arranged, integrated, and made
available in data systems and within the organization. Be
able to manage and use external data sources,
including an API layer within your data
Get the most from your data
Make good use of data visualisation
techniques and take pains to highlight and
display key information in a user-friendly way
to help ensure that your data gets to good
use. Define how the insights will be
communicated to the information consumer
or decision maker. Consider whether
interactivity is a requirement, if key decision makers
in your business need access to interactive self-
service reports and dashboards.
"Machine learning allows us to
build software solutions that
exceed human understanding
and shows us how AI can
innervate every industry."
-Steve Jurvetson,
Board Member of SpaceX and Tesla
Prepare the employees
Be transparent with your employees, informing them on the organisation strategy,
clarifying your willingness to the digital transformation and to AI adoption. Plan an
adequate training and information days to prepare them for the digital
transformation. Communicate the vision and the value of data.
Focus on value-added activities
Demonstrate how AI can remove the burden of repetitive and non-value-added
tasks or to contribute to human decision making, allowing employees to focus on
more creative and fulfilling work, learn new skills and enjoy a more varied career.
Celebrate the success stories
Celebrate insights generated through data and success stories achieved through
data analysis (e.g. by means of Business Intelligence solutions or monitoring
Cover skills gap and train your employees
Assess employees’ capabilities and skills gap in order
to be aware of owned competences. Plan for
each of your employee a specific training path
considering company’s digital journey and
their future prospects. Don’t forget to involve
everyone inside the company, in order to allow
them to actively learn and be prepared to the
new solutions.
Review the role in the company
Review your organization’s structure to identify if
dedicated positions to support AI adoption, such as
Data Engineers, Data Scientists, are part of the
organisation blueprint.
Identify new roles
Define data and AI-related roles (e.g. data Scientist, data Engineer).
Continuously train and certify employees in these positions in order to be
aware of latest developments and innovation in the field of AI.
As more and more
artificial intelligence is
entering into the world,
more and more
emotional intelligence
must enter into
- Amit Ray,
AI Scientist
Find a talent pipeline
Promote active talent acquisition of AI-enabled staff as well as talent management
of the standing organization.
Raise the work
Drive human workers to display greater creativity, objectivity and agility to
maximise the benefits of AI and drive the organisation forward.
Promote innovation
Allocate a yearly budget to digital innovation initiatives considering that disruptive
technologies could rapidly change the game and the money spent could be
rapidly recovered.
Restructure the roles in the organization
Rethink the roles of each worker inside the organisation designing assignment of the
new roles that will born to the employees with the right competencies. Rely on the
assessment of the capabilities of the workers and their willingness to be involved in
this process of change.
Integrate AI in the company
Integrate AI-driven roles in an agile organizational structure.
Promote an innovative culture
Promote a digital, data and AI-driven culture favouring new ways of thinking and
the capacity to re-invent organisation’s roles with the introduction of new
technologies. Since employees are at the centre of the disruption, they should have
the time to learn new ways of doing their tasks and improve how these are carried
Increase trust the data
Build trust in data, algorithms and AI by realising simple use-cases demonstrating
their effectiveness and their capability to solve employees’ problems. Simple
projects with a small but well-defined scope projects are in many cases more
effective than complex and long ones.
Spread how AI transform the daily work
Maintain a steady stream of positive news about how the AI is helping individuals,
teams and the organisation as a whole to be more successful. Be transparent about
both successfully and non-successfully projects.
Increase data awareness
Achieve a high level of awareness regarding data-related issues in employees to
prevent future cases of the “garbage in, garbage out” syndrome.
Deal with ethics
Form a team dealing with ethics of your AI solutions.
Much of what we do with machine
learning happens beneath the surface.
Machine learning drives our algorithms for
demand forecasting, product search
ranking, product and deals
recommendations, merchandising
placements, fraud detection, translations,
and much more. Though less visible, much of
the impact of machine learning will be of this
type quietly but meaningfully improving
core operations.”
- Jeff Bezos,
Founder and CEO of Amazon
The Sustainable Production Systems Lab (SPS) is a research
institute belonging to the University of Applied Science and Arts
of Southern Switzerland (SUPSI). The mission of the Institute is the
innovation of production processes and business models in
order to supporting companies in facing the challenges of
digitalization under the economic, environmental and social aspects. The fulfilment
of the mission is achieved through the development and technology transfer
activities with reference to the life cycle of products and industrial processes, in the
fields of design, automation and management of production systems.
This research has been carried out in collaboration with
Warsaw University of Technology (WUT) is the leading technical
university in Poland. WUT has an established network of
international collaborations, including over 150 recognized
universities worldwide and many global corporations.
Ginkgo Analytics is a spin-off of Ginkgo management consulting
and is based in Hamburg and Munich. Our mission is to transform
companies into data-driven and AI-powered enterprises.
CRIT is a SME innovation company belonging to 26 large
manufacturing and processing industries in Emilia-Romagna. CRIT
addresses services both to shareholders, mostly large companies,
and to their suppliers, which are mostly SMEs.
ART-ER Attractiveness Research Territory is the Emilia-Romagna
Joint Stock Consortium born with the purpose of fostering the
region’s sustainable growth by developing innovation and
knowledge, attractiveness and internationalisation of the region
Innovation Centre Nikola Tesla is the leading constituent of the
innovation ecosystem in Croatia, and Danube region in
perspective, for applied research and development in the field of
engineering and related applications.
The Spanish Council for Scientific Research (CSIC) is the largest
public multidisciplinary research organization in Spain. It has a staff
of more than 10000 employees, among these more than 3200
scientists and about 3800 pre- and postdoctoral researchers.
Our team
Andrea Bettoni
Senior lecturer and researcher
Donatella Corti
Senior lecturer and researcher
Zeki Barut
Scientific Collaborator at
Elias Montini
Researcher at SUPSI-SPS Lab
Davide Matteri
Assistant researcher at SUPSI-
Sara Masiero
Assistant researcher at SUPSI-
Michele Fiorello
Scientific Collaborator at
Sabine Gretenkord
Data scientist at Ginkgo
Bartłomiej Gładysz
Research and teaching
assistant professor at WUT
Krzysztof Ejsmont
Research and teaching
assistant professor at WUT
More details on our AI maturity and adoption model have been published
within the IFAC-INCOM 2021 Conference paper
“An AI adoption model for SMEs: a conceptual framework”
ResearchGate has not been able to resolve any citations for this publication.
ResearchGate has not been able to resolve any references for this publication.