Like many large corporates Bosch, a German multinational engineering and technology company, is interested in putting AI into production to make its factories more productive and competitive. Although the machines and production line in its plants are already well-equipped, integrating new sensors and deploying AI technology could further enhance efficiency. While the company has an open innovation department and regularly works with startups and scale ups, its Maklar, Hungary plant had little experience in working with young tech companies.
Through the DeepTech Alliance, a private non-profit association of leading European entrepreneurship hubs that specializes in connecting deep tech companies with large corporates, the innovation manager of Bosch’s plant in Maklar connected with IPercept, a Swedish startup that serves as a kind of fitness tracker for industrial machines, leveraging AI to track mechanical movements to do root cause analysis and predictive maintenance.
Bosch tested the technology on a welding machine used in the automotive steering systems it produces. The objective was not just to tackle unplanned downtime, but to preempt it, leading to reduced costs, improvement in product quality, and elevated customer satisfaction. The pilot was a success, and the Maklar factory is now looking at how to apply the technology to other production lines, says Gabriel Gudra, the factory’s innovation manager. He and IPercent CEO Karoly Szipka appeared on stage together to talk about their collaboration at an October 9-10 DeepTech Alliance advanced manufacturing meeting in Munich which brought corporates together with startups targeting the manufacturing sector.
Bosch’s story is just one example of how manufacturers are starting to unlock AI’s potential. The same week the DeepTech Alliance was hosting the advanced manufacturing event in Munich the World Economic Forum announced the latest additions to its Lighthouse network, a community of 172 industry leaders pioneering the use of cutting-edge technologies in manufacturing. Nineteen manufacturing sites received designations as Fourth Industrial Revolution Lighthouses for achieving step-change impact in performance through technology-enabled transformation. Three others were designed as “Sustainability Lighthouses” for their use of advanced technologies to reduce their environmental impact.
The Forum’s latest cohort gained an average 50% boost in labor productivity, attributed to various digital solutions such as interactive training programs, smart devices and wearables, and automated systems that combine robotics, AI and machine vision. Process modelling and root-cause analytics unlocked efficiency gains across Lighthouses’ end-to-end supply chains, on average reducing energy consumption by 22%, inventory by 27%, and scrap or waste by 55%, according to the Forum.
“There are not many examples of successful applications of AI in industry,” says Federico Torti, the Forum’s Initiatives Lead, Advanced Manufacturing and Value Chains. “The lighthouses are demonstrating what can be achieved and serve as an inspiration for other companies.”
What sets the lighthouse companies apart is that they have already made critical investments in their tech stacks, and they design AI use cases for scale, creating an easily replicable package across their production network, says Torti. They start by fundamentally redesigning processes, reducing variability and waste, and organize technology deployment around the user, focusing on people skills and user experience / how users interact.
The lighthouse companies, which hail from 10 different countries, are also open to learning from their peers and across sector, he says. “They are really pushing the boundaries of how they can get value from cutting-edge technologies like AI,” says Torti. “It is not about piloting AI, it is about creating the right ecosystem that will allow their company to transform.”
In general, these companies are taking a long-term oriented approach, have a vision and are investing in “the required foundational aspects – such as clean, reliable data platforms, a good governance structure and training their work forces with a people-oriented approach –rather than trying to quickly adopt the shiny aspects of AI,” he says.
Global pharmaceutical company AstraZeneca’s Södertälje plant in Sweden – one of the Lighthouse factories in the Forum’s new cohort – is a case in point. The plant has implemented 50+ advanced technology solutions and a significant number incorporate AI or GenAI, Jim Fox, VP Sweden Operations, AstraZeneca, said in written responses to questions from The Innovator.
In drug development, AI predictive modeling is optimizing the physical and chemical properties of Active Pharmaceutical Ingredients (API) and predicting the performance of formulated products during manufacturing. What’s more GenAI, machine learning, and large language models (LLMs) are significantly reducing development lead times and the use of API in experiments, says Fox. AI-powered process digital twins are optimizing conditions for yield and productivity in the manufacturing process, reducing the use of raw materials, and minimizing tech transfer requirements. AI-powered tools are also aiding AstraZeneca in achieving its Net-Zero carbon footprint goal by pinpointing its environmental emission “hotspots” and the carbon footprint of itsd products across the entire supply chain.
An important factor in its success to date was preparing “findable, accessible, interoperable, reuseable and clean data” to power its AI algorithms and ensuring it is managed with good governance and data standards, says Fox.
Upskilling 3,000 employees to deal with VR/AR, AI computer vision, IoT, sensors, integration of cobots, drones and digital twins was also key. AstraZeneca has a strategic workforce plan that includes both outsourcing/offshoring AI services and building critical internal capabilities to cover the range of its needs, says Fox.
Specific tailored training for employees at the Swedish site was introduced through an extensive online training platform. “We also launched a Digital Academy together with a digital innovation zone for experimentation to sustain our digital capability uplift,” says Fox. In addition, the plant has also launched a comprehensive and accredited self-serving online program for basic to advanced AI and Gen AI training.
“The experience at the Södertälje plant has provided valuable insights into scaling AI and upskilling employees effectively,” he says. “If we had not done [the upskilling] we would not have seen the results that have led to a 56% increase in production and a 67% reduction in development lead times for launching new products.”
The Human Factor
The human factor must be considered in any successful adoption of new technologies, says Antti Rantanen, who runs the industry 4.0 and Industrial AI practice of the Nordics division of EFESO, an international consulting group specialized in helping manufacturers to use technology to advance operations strategy and performance improvement.
He cites a brewery’s manufacturing plant in northern Europe as an example. Most of the employees have been there for 30 to 35 years. “We were walking around on a factory tour when a bottle machine got stuck and the 60-year-old operator knew exactly where to kick it to get it started again,” says Rantanen, a speaker at the Deep Tech Alliance Munich event. An estimated 30% of the work force in Nordic factories will retire in next 10 years, he says, and that knowledge will disappear. Top-down technology solutions imposed by management are not the right fix.
“Factory operators hate it when headquarters comes and visits,” he says. “They have their own way of doing things.” Take the case of one packaging company Rantanen visited. The company’s leadership installed new SAP software. When leadership visited shop floor operators would open laptops to something that looked like the German tech company’s product and repeatedly push their shift buttons to make it look like they were using it. As soon as management left, they would go back to business as usual. “They explained to us that using a complicated IT system does not help them do what they need to do and if they adopted the technology it would ruin their P&L,” says Rantanen, who has 25 years of experience in digital transformation. “This is the situation in most manufacturing plants.”
When headquarters imposes AI solutions or software “they rarely ask the machine operators ‘what do you see as the inefficiencies and how will this connect with the overall value chain?’,” says Rantanen.
EFESO helps companies take a different approach. “We walk through a production plant before we start any process to understand the culture of that plant,” he says. “We usually meet with the CEO, plant manager, safety and maintenance – all the way down to the machine operators – to understand how people work together. We earn the trust of the machine operators, then, we go at this from an operational excellence perspective, find the inefficiencies and build the AI models with partners like SILO to be scaled out.”
Success depends not only on the input of people using the technology, with deep knowledge of operations. Cultural differences must also be considered, he says.
The French manufacturing culture is different from the German or the Swedish, says Rantanen. “Global companies think that they can roll out the same systems and same architectures in the same way but if you look at a plant in, say, Finland versus one in Brazil, the plants will be completely different in the way they work, their openness to change, in whether they adhere to processes or not,” says Rantenen. “Those things need to be taken into account, not just AI or robotics.” These issues will be ongoing even after AI is implemented and factory jobs change, he says. “AI will be essential for manufacturing companies to stay competitive globally,” says Rantanen. “But it will never work if they don’t take the human elements into account.”
Scaling Up
With all the hype around Machine Learning and AI “we get the impression that it is all changing very quickly,” says Thomas Klem Andersen, Executive Director of the DeepTech Alliance. “But a lot of manufacturers are still not there,” he says. “A lot more has to be done not just for the bigger industrial companies but also the smaller ones to help connect them to the right deep tech startups, identify use cases and implement them.”
This article is content that would normally only be available to subscribers. Sign up for a four-week free trial to see what you have been missing.