More than 5,000 players from the Canadian manufacturing industry gathered on November 9 and 10 for the Advanced Design & Manufacturing 2022 event at the Palais des congrès de Montréal.
As part of this important gathering, Scale AI’s CEO Julien Billot held a panel on artificial intelligence, alongside key stakeholders from Kruger, Pratt & Whitney Canada, MDA and IVADO Labs to talk about the challenges and opportunities they encountered in their journey with AI.
“Implementing AI is more a journey than a destination in and of itself. This journey includes several steps that can vary from one business to the next, but one thing is certain: AI is here to stay. Organizations that understand this and that already fully integrate AI into their strategy will be ahead of the competition in the future.”
— Julien Billot, CEO, Scale AI
Start from the beginning: What is your objective?
Jack Klejka, VP Product at IVADO Labs, says it’s important to start with a specific use case, with a clear goal in mind. Whether your goal is to improve workforce planning, to get visibility on your inventory or to reduce downtime in your production line, the first step of any AI project should be to define its objective.
The first AI project developed at Kruger was built for their brand new state-of-the-art plant located in Sherbrooke, with a clear objective in mind.
“Our goal was to increase productivity at this facility. We started with a specific use case and saw promising results early on in the project. To date, we have 8 use cases in our Estrie plant and we will continue our exploration of AI with several other projects planned for 2023.”
— Heather Scott, Director of OPEX Business Solutions at Kruger Products
Assess: Is your data actionable?
To ensure the success of your AI project, it is essential to assess upstream the quantity and quality of the data that is currently being collected within your organization, and to determine if this data can be used to build the algorithms that will allow you to achieve your initial goal. In the event that you need data that is currently not being collected, it is possible to be creative. Isabelle Gailloux, Senior Manager, Supply Chain & Production Control at MDA Space: “On our side, we used existing data, but we also created dummy data as placeholders to build our algorithms. This allowed us to develop our models, while clearly identifying the data that we would need to collect in the near future so that our algorithms help us achieve our objectives.”
Compute: The power of algorithms
AI allows manufacturing organizations to save time and energy, thanks to algorithms that can take into account a very large amount of data to identify optimal solutions. Heather Scott, Director of OPEX Business Solutions at Kruger Products: “A production line at Kruger can involve five to ten thousand settings. The possible combinations to optimize productivity are therefore numerous. Having a process engineer try to find the best parameters is a full-time job, for more than one person. Thanks to our algorithms, we are able to define the optimal setting for each parameter so that we achieve our goal of maximizing our production.”
Predict: Better plan the future
Pratt & Whitney has used AI to predict spare parts demand for its 65,000 engines in service in 200 countries around the world. Sébastien Turcotte, Associate Director, Artificial Intelligence Transformation Lead at Pratt & Whitney Canada: “The goal was to improve the forecasting of the consumption of all spare parts, in order to be able to optimize our inventory level. Thanks to what we have put in place, we are able to guarantee that the right parts are in the right location at the right time, which saves our team a considerable amount of time while reducing waiting times for our customers.”
“Demand forecasting is the gateway to AI. This is the most accessible way to start your journey in artificial intelligence. The benefits of projects related to predictability are quite convincing; every member of your organization can easily understand how those initiatives can have a direct impact on improving margins and increasing the company’s turnover.”
– Jack Klejka, VP Product at IVADO Labs
Automate: Leveraging predictive models
With predictive models in place, organizations can plan to automate certain recurring tasks, or even entire processes. In the case of inventory management, for example, one can consider automating the ordering of a given material when the stock reaches a certain predefined level. This will avoid any stock shortages.
