Journey to AI – Use Case

Publication date

In today’s era of data-driven technology, businesses across all sectors are scrambling to put artificial intelligence to work for them. But venturing onto unknown territory is both exciting and daunting, and the requisite learning curve can feel like a mad rush and slow going at the same time. From C-level executives to seasoned IT professionals, everyone is saying the same thing: it’s hard to know where to start, but there’s no time to waste!

R2i helps businesses of all sizes navigate the journey to artificial intelligence. We recently worked with a mid-size Canadian manufacturer to develop their first AI project. Here’s a look at what we did.

The initial conversation

The company’s CEO wanted to know how AI is redefining the manufacturing field and how he could make sure his business wouldn’t get left behind. We started by providing an overview of how local competitors and industry leaders are using AI in a wide range of areas like preventive maintenance, quality control and demand forecasting.

Our client hoped this quick overview would make it immediately clear how his company might integrate machine learning into its processes, but establishing your first AI use case isn’t as simple as that. We explained that the key to success lies in a carefully selecting a business problem that’s specific, measurable, achievable and—importantly—backed by large volumes of existing data.

The client pain points

Our next step was to lead the company through a brainstorming session to explore the company’s various pain points and shortlist the ones that showed the best potential for success through AI. We learned that the company had a serious need to improve its demand forecasting because changing market conditions (like small batch orders and the need for faster deliveries) put tremendous pressure on lead times and the manufacturer faced costly penalties every time it was late in fulfilling orders.

While the stakes were high, even small gains in forecasting accuracy would lead to major financial savings, not to mention more balanced inventory levels, improved customer satisfaction and increased sales.

The process

Once we’d agreed on this compelling use case and ensured that the company had sufficient available data to enable machine learning, we went to work defining the criteria for assessing a minimum viable product, which in this case would be an AI-based demand forecasting model for a single product category. A narrow scope would allow us to reduce the number of variables and achieve quick wins (or fast fails, depending on how you look at it).

To stack the cards in their favour, the client leveraged IBM Watson Studio, a leading predictive analytics and machine learning solution to build the minimal viable product.[1] This product would allow different project stakeholders (data scientists, application developers and subject matter experts) to collaborate and train models at scale, and to deploy in a hybrid environment for faster operationalization. The project also leveraged R2i’s private cloud, which is backed by ISO 27001 and 27018 certifications for enhanced information security and personal data protection.

R2i has spent the last few years navigating the Canadian AI ecosystem. As a result, we led the application for funding from ScaleAI, one of Canada’s five superclusters. The application was submitted by the manufacturer, R2i and one other company—the three stakeholders that would jointly operationalize the solution–and once it was accepted, the project received funding for 50% of the total cost.

Finally, we connected the client with Mitacs, an organization that helps industrial companies access leading technological expertise at a reasonable price through collaborations with post-secondary institutions.

The results

After three months of testing different models and data sources, we finally landed on a model that led to a 5% improvement in the company’s predictive analytics, compared with their non-AI legacy system. This gain was achieved using just one additional data source and AI. Not only was the improvement appreciable in and of itself, it also set the foundation for future AI projects targeting the company’s other product categories.

The client is now looking to build on this achievement by using additional data sources and fine-tuning their AI models. We are continuing to assist them with technical aspects, including hardware management, and with further applications for funding to enable the mid-sized manufacturer to modernize its operations and operationalize AI.  In order for their business to survive, they need a forecasting model that can adapt to quickly changing market demands. In the era of Industry 4.0, it’s a new reality that all manufacturers have to face—and that all can successfully manage with the right tools and partners.

[1] Forrester, Q3 2018 Wave Report on Multimodal Predictive Analytics and Machine Learning Solutions

R2i, experts in Artificial Intelligence

Whether you are considering investing in hardware and cloud computing solutions, strategic support from a trusted partner like R2i provides technical and strategic support to organizations of all sizes that are preparing to adopt artificial intelligence. As facilitators, we can help your company take the 4.0 turn, develop a solid business case, grow in maturity and put your game plan into action with brio. To find out more about the link between artificial intelligence and cloud computing services, find out now what artificial intelligence can bring to your business.

You can also contact a member of the R2i team for any additional information and to learn more about our consultants’ expertise and know-how.

Share on your social media