AI in manufacturing

Navigating the challenges and opportunities.

Alessandro Di Carlo

The rapid adoption of artificial intelligence (AI) marks a significant shift within the manufacturing industry, redefining production, efficiency and process innovation. It is not just a technological breakthrough, but a major revolution that transforms the way products and services are created. With its multiple applications, AI addresses longstanding industry problems, converting technological potential into practical solutions that significantly impact business operations and strategy.

One of the key issues addressed by AI is inefficiency in production and in technical and commercial back-office processes. In today’s global market, where resource use is critical, AI’s optimisation capabilities can make a substantial difference and act as a powerful business accelerator. It enables real-time analysis of extensive numerical and textual data, making logical connections and offering insights to pinpoint inefficiencies and suggest enhancements. This reduces errors and increases productivity. AI also significantly reduces human errors, which often lead to waste and delays. Systems that continuously learn and improve execution accuracy and speed make efficiency the norm rather than the exception.

The manufacturing sector, traditionally slow to adopt new technologies, is now undergoing major changes. Companies that collaborate with partners to help them identify the potential areas of application of AI according to their level of maturity are successfully navigating the challenges of an increasingly competitive market. AI is no longer just a future possibility but a real, accessible opportunity, an investment in progress that is poised to revolutionise the manufacturing landscape.

The state of the art in the machinery manufacturing industry

In any technological evolution, an information overload can confuse non-experts, making it hard for them to evaluate the many options on the market. This can leave them undecided or lead them to make poor decisions that slow down adoption. Language interpretation is one of the most challenging tasks for a machine. There are several methods for understanding and analysing text, including symbolic approaches, Large Language Models (LLMs) and Generative AI, all of which can yield satisfactory results depending on the specific needs and contexts. Experimenting and prototyping with AI is one thing, but using it in daily production activities is quite another.

To optimise a process, deploy it in production and make it scalable, a hybrid approach that combines different AI text analysis techniques may prove to be the best choice. It is essential to concentrate on the process, key performance indicators (KPIs) and objectives, leaving it to AI experts to choose the right toolset to achieve the desired results.

Company managers should take account of the experience and field knowledge of their AI consulting partners and consider solutions already adopted by other companies to understand their advantages and limitations. To integrate AI into business processes, it is necessary to form a capable, collaborative and motivated team, with a team leader who can liaise effectively between the internal team and the AI consulting partner. These steps are essential for pursuing successful projects and clearly demonstrate how artificial intelligence can effectively support specific activities and avoid wasting time and money on low-value trials.

Begin with the build-up approach to AI

Transitioning from the concept of integrating AI into a business to implementing it in a specific use case is rarely straightforward or intuitive. Furthermore, leveraging AI to generate business value demands significantly more focus than merely experimenting with the many tools available on the market.

Tools evolve and newer, more efficient technologies are constantly emerging. Today’s best tools are likely to be rendered obsolete by tomorrow’s innovations. This makes it essential for a company to adopt the right combination of methods, technical skills and cultural approaches to maintain the best trajectory even as tools evolve.

The build-up approach is a specific method for introducing AI into a company based on consultancy, allowing for immediate testing of strategic applications while simultaneously identifying gaps that need to be filled to enhance the integration of AI across the various processes.

This approach is based on objective analysis, experience and the consultant’s technical and operational vision. Only then does the mix of products and technologies needed to achieve the goals come into play. Essentially, the build-up approach is a consultative process consisting of specific, clearly defined phases, from equipment analysis and the identification of opportunities to pilot projects, all focused on continuous improvement through regular review.

Phase 1: AI equipment analysis.

Specific tools and checklists are used to analyse the data available within the organisation, thereby determining what data exists and its accessibility. This initial phase helps to identify areas for improvements in data collection, highlight process issues and define the concept of data quality for a specific process. Companies often amass vast amounts of data without a specific purpose. This raises important questions: Who is responsible for defining data quality within the company? When can data truly be considered high-quality? Data quality is not necessarily determined solely by precision and accuracy. There are many different aspects that need to be considered to accurately define quality for each process to be optimised.

Essentially, the correct data to be collected depends on the questions that are to be answered. This vital concept is frequently overlooked, resulting in indiscriminate data collection without a clear purpose. The goal of the AI Equipment Analysis phase is to map the available data, their characteristics and the gaps that need to be filled.

Phase 2: evaluation of the best AI opportunities.

The second step involves identifying the potentially most valuable AI initiatives. These opportunities might arise from available data sources or from known inefficiencies in specific processes. After analysing these processes and inefficiencies, the consultant must pinpoint areas where AI can be immediately useful and resolve issues. The aim of Phase 2 is therefore to identify the area or process that needs to be optimised and evaluate its feasibility and potential impact.

Phase 3: pilot project and optimisation.

Phase 3 consists of two related activities: transitioning to operational implementation and beginning to address data collection inefficiencies. This phase is critical and should be completed within a maximum timeframe of 4-6 months.

Pilot Project - Once the area of intervention (such as customer service) has been identified, the scope of a specific AI use case can be defined. This involves performing a detailed analysis, planning the necessary activities and implementing a pilot project that delivers immediate value, avoiding mere stylistic exercises.

Optimisation - The company and consultant will choose which inefficiency to focus on in order to refine the data collection method and gradually fill the gaps, allowing for the continued optimisation of all processes that can benefit from AI.

Objective - Within the first 4-6 months, the company will have integrated an AI-optimised process into its daily workflow. Tools for tracking initial key performance indicators (KPIs) will be available right from the outset, along with everything needed to collect end-user feedback. Additionally, improvements in data collection will have been implemented, placing the company in an ideal position to identify new areas for optimisation.

Phase 4: review and improvement. The review phase enables the company to evaluate the effectiveness of the deployed technological solution. In this period, the company can assess the artificial intelligence model’s performance, gather feedback from operators and monitor user interactions with the new system. This allows it to measure user satisfaction and understand the system’s impact on daily operations.

The advantages of the build-up approach

In the contemporary business landscape, integrating artificial intelligence is essential for maintaining competitiveness and innovation. However, a strategic and structured approach is needed to support this shift effectively. The build-up approach stands out for its ability to align AI usage with strategic business objectives, promoting a culture of ongoing innovation and agile adaptation. This method involves skill acquisition, careful financial management and the development of a thorough understanding of AI across all levels of the organisation, helping to maximise the return on technology investments and minimise associated risks.

Below, we explore in detail how the build-up approach translates into concrete benefits for companies that choose to adopt it, improving not only operational efficiency but also the ability to proactively respond to market challenges.

Data strategy vision. Defining a clear data management strategy is essential in order to optimise processes. Many firms have a solid business strategy but lack an equivalent data strategy, which is crucial for supporting overall business goals. In order to support more informed decision-making and strategic actions, data quality should be evaluated not only for its precision but also in terms of its relevance and efficacy within the business context.

Gradual learning. AI adoption requires a constant commitment to staff upskilling. Ongoing education in the latest technologies and best practices not only improves the ability to work effectively with the new technologies but also helps to reduce resistance to change, creating a more dynamic and innovative workplace. Training should be viewed as a continuous investment in the workforce, enhancing the organisation’s ability to adapt and respond to new challenges.

Distributed budget allocation. Proper budget allocation is crucial to successful AI integration. Investments should be strategically distributed to cover not only technologies and infrastructure but also employee training and external consultancy. Financial planning must be sufficiently flexible to adapt to the evolving needs of the project and new insights that emerge during the implementation phase. Proper budget management ensures that resources are used in the most effective way possible in order to drive innovation and maintain competitiveness.

Widespread diffusion of AI knowledge throughout the company. An in-depth understanding of AI within the company is vital, not only for the IT team or technicians but for all levels of the organisation. Awareness of how AI can influence and improve various aspects of the business can spark new ideas and innovative applications. This shared understanding facilitates greater collaboration and acceptance of AI-based solutions, integrating such technologies into daily business practices.

Alessandro Di Carlo, Expert.ai (Modena, Italy)