The retailer is looking to embed artificial intelligence across the business
While every major enterprise is undertaking the implementation of AI projects to one degree or another, the big wins will be for those that scale.
The reality is that significant results may require scaling. We’ve heard plenty of examples of this at the recent virtual AI Business Week Digital Symposium.
For example, Babylon Health described how it was using artificial intelligence to answer medical questions from consumers via mobile devices. With a ratio of 11 million people to 1,000 doctors in Rwanda, Babylon’s automated system found a suitable home, according to Umang Patel, the company’s managing director for value-based care. In this case, the Babylon system is most effective when there’s a market of a considerable scale.
In the case of Branded Entertainment Network (BEN), AI plays a role in numerous parts of the business, ranging from determining the most appropriate products to be included in a film or a TV show, to predicting conversion rates. In this case, scaling up meant using AI to analyze millions of items of unstructured data, said Ricky Ray Butler, CEO of BEN.
In its approach to scale, H&M saw an opportunity and necessity to evolve, realizing that no fashion player had taken a lead position in AI. H&M started with AI at the group level in 2016, and quickly evolved from there. By 2018, the clothier had set up a totally new function, providing for its own budget and investing in staff with an AI focus. When the global pandemic hit, the company decided to increase the momentum.
“In the middle of our great plan, COVID kicked in,” Errol Koolmeister, product area lead engineer for AI Foundation at H&M, explained at the AI Business Week event. “We decided we really needed to push forward with everything,”
The vision became “to make H&M Group the industry leader in applied AI with scalable and integrated solutions covering the entire value chain.”
H&M created the Fountainhead, a platform focused on “democratizing” AI within the company, with the intent to be able to work with “hundreds, if not thousands” of use cases throughout the enterprise.
“We wanted to make AI available across the entire H&M group,” Koolmeister said.
To take AI to the next level, the AI team wanted to become a facilitator for scalability. “For us, it was mainly about creating foundational teams that facilitate the scale of AI capabilities,” Koolmeister said. “Rather than going by use case to use case, we focus on training and development.”
Foundational capabilities include knowledge capture and management, training and development, exploration and research, quality assurance, AI risks and ethics, expertise, data science tools, and ecosystem and vendor management.
The objectives were to reduce time-to-market for AI use cases, make AI tools available for all product teams, contribute to a higher skillset in business tech via AI literacy and hands-on support, and increase model output availability, such as demand forecasts, for all.
The idea was to streamline AI development and share the capabilities throughout the enterprise with the added benefit of avoiding a wrong path – Koolmeister detailed what can go wrong when trying to get an AI system into production, and the process H&M used to get it right, in the VisionAIres community.
He said that the framework has delivered business value by reducing time-to-market for use case development by 50 percent, from 12 to six months. It also unlocked use case value potential across the entire value chain, and fostered a co-creation and sharing culture.
Considering the success of the model, H&M seems to be pointing the way to the next level of AI adoption in an enterprise.