The industry has always faced new challenges brought by technological evolution, and today, following the ongoing wave of digitization, it is the era of artificial intelligence and GenAI, or generative artificial intelligence.

With its rapid development curve and widespread adoption, GenAI is already revolutionizing the world of artificial intelligence. Its advanced information processing models are accelerating the transformation of machine and plant data into actionable insights—just as promised by Industry 4.0—overcoming numerous complexities.

The world of machine builders and that of manufacturing companies using industrial machinery both need guidance: one towards new business models, and the other towards the implementation of increasingly efficient and sustainable production processes.

This is the goal of 40Factory, a cutting-edge scale-up from Piacenza specializing in AI-IIoT industrial solutions. With its innovative offerings such as MAT, the white-label Industrial IoT solution, and Wilson.ai, the Generative AI-based virtual assistant, 40Factory aims to support the manufacturing world in this new transformation. Fabio Vesperini, Head of Data Science, guided us through the world of 40Factory.

The new challenges of the industry according to 40Factory

With the advent of Industry 4.0, companies have gained the ability to collect data from their machines. Transforming this vast amount of data into actionable insights to optimize efficiency and maintain competitiveness represents an additional challenge.

In 2018, Camillo Ghelfi, the current CEO, founded 40Factory with the goal of developing an IIoT platform to assist machinery players in navigating a radical shift in their business. This shift involves complementing traditional machine sales and construction with new approaches and business models.

Over the years, thanks to strategic investments and rapid revenue growth, the company has expanded significantly. Today, 40Factory employs around twenty young professionals and operates across Italy with more than 40 agents. It also has a commercial presence in Germany and collaborates with various companies in different sectors of the machinery industry, including Food & Beverage, machine tools, and textiles.

From the IIoT MAT Platform to Wilson.ai’s GenAI

40Factory has grown alongside and around its IIoT platform, MAT (Machine Analytics Tool), which enables real-time data acquisition from machines or entire industrial plants. This data is then used to provide detailed analysis on various aspects such as productivity, performance, and consumption, allowing decisions to be made to optimize and maximize the efficiency, productivity, and sustainability of production systems.

“MAT is a modular system and was designed as a white-label platform, offering maximum flexibility for configuration and customization,” explains Fabio Vesperini. “In addition to real-time data monitoring, modules can be added for monitoring specific components or modeling particular production processes that require dedicated machine learning and data analysis algorithms, all based on the customer’s needs.”

The platform is aimed at machinery and industrial plant builders, who can also benefit from a range of additional services enabled by this technology, such as service support, technical assistance, and even a “pay-per-use” model—meaning the use of equipment under performance-based contracts rather than the sale of physical assets accompanied by a service contract.

“MAT continues to evolve with our client base, a community that enriches the platform with new features based on field experiences, including those from different contexts,” adds Vesperini. “The latest development of MAT is the integration of GenAI, with Wilson.ai.”

What Wilson.ai is and what it does

Wilson.ai is a generative artificial intelligence-based virtual assistant that continues to evolve and gain new functionalities. “Wilson.ai was created to provide technical support, aiming to alleviate the workload of technicians, particularly for frequent and repetitive interventions,” Vesperini explains.

“Machinery and plant builders produce a lot of documentation to accompany their machines, such as manuals, guides, and schematics, which are often difficult to consult, especially when interventions are required in specific production contexts. Tools like Wilson.ai, based on Large Language Models (LLMs)—an advanced AI technology focused on text comprehension and analysis—are radically changing the way interventions are performed on industrial plants, assisting and relieving technicians in their maintenance tasks.

But that’s not all. Like all intelligent tools, Wilson.ai continues to learn and expand its knowledge and skills, also learning to solve frequent problems that arise, for instance, during machine commissioning or related to specific faults. This way, such expertise is not limited to the personal knowledge of technicians but is somewhat digitized, consolidating and enhancing the company’s know-how and keeping it within the organization.

Wilson.ai is easily accessible to anyone, communicating in natural language and multiple languages. Machinery builders often install their machines around the world. Imagine being able to solve technical issues simply by consulting a virtual assistant. Thanks to its ease of interaction, Wilson.ai can even be used by the machine’s end user.

Beyond technical assistance with Wilson.ai

Beyond the realm of technical support, Wilson.ai can also be used in other contexts. For example, it can assist anyone needing to query various types of documents such as contracts and company procedures.

Among the new features, we find automatic image extraction and classification, text-to-speech and speech-to-text capabilities, and integration with MAT, which ensures a continuously updated flow of information. Thanks to the power of GenAI, Wilson.ai can query MAT and retrieve valuable information.

In this way, Wilson.ai not only performs data analysis and correlations based on observed data but also acts as an expert user of MAT. It can respond to queries posed in natural language, providing answers in both natural language and through dashboards, tables, and charts, thereby accelerating the use of MAT and making the information more accessible.

Secure data within companies

Data security and ownership when dealing with generative artificial intelligence models are key concerns for businesses. For 40Factory, digital data security is a top priority. The architecture of its solutions is designed on inherently secure technologies and infrastructures, ensuring complete separation of machine and client data, with a certified and traceable access control system.

In addition to managing and storing the information and documents used to train Wilson.ai with state-of-the-art cloud security systems, 40Factory also offers clients the option to retain data ownership and storage in-house.

“Currently, we are using OpenAI models available on Azure,” explains Vesperini. “About two years ago, we were selected by OpenAI to access these models’ APIs for developing GenAI solutions. One important requirement is that the data exchanged and used to generate responses for solutions like Wilson.ai will not be used to train the generic models.”

40Factory’s experts are also working on developing local GenAI models, which can operate with reduced computational resources—without needing to rely on the cloud—while maintaining the same cognitive capabilities.

These models can thus be used on local servers and PCs in an edge computing framework, containing the data within the physical boundaries of the client company. The evolution of LLMs is moving precisely in this direction.

Wilson.ai in action: two Case Studies

MFL Group is a company that manufactures machinery for the production of cables and ropes. In such a setting, the production process is crucial, as the machinery is custom-developed for each type of production. In this context, having access to a repository of company know-how is highly valuable for retrieving information when issues arise, especially for recurring interventions that can be resolved by a virtual assistant like Wilson.ai, significantly easing the service workload.

Another company currently experimenting with Wilson.ai is Fincantieri, a leading Italian shipbuilding company with one of the most significant shipyard complexes in the world. Fincantieri employs approximately 19,000 people, operates 20 facilities across 4 continents, and has commercial relationships with major cruise operators, the Italian Navy, and even the US Navy. The collaboration, which began in December 2023, is already yielding impressive results.

We are just at the beginning of the new era of Generative AI.

Original Source: Automazione News