How Howdens used GenAI to power app user experience

Howdens, which supplies kitchens to tradespeople, has developed a new artificial intelligence (AI)-powered app to provide technical knowledge. Chip uses Microsoft’s generative AI (GenAI) technologies, including Azure OpenAI Service, Azure Cosmos DB, Azure AI Search and Applied AI Services.

Chip has been designed to provide Howdens’ trade customers with the information they need to do the best possible job, all in one app.

Howdens is an existing Microsoft customer, and as David Sturdee, chief customer officer at Howdens, explains, GenAI opens up a number of opportunities. “With the advent of GenAI, we immediately knew that this changes everything,” he says.

While some of the opportunities resulting from the use of GenAI may be small now, the direction of travel means that in five years, these small changes have the potential to change how the business operates. “What we identified quite early on is that GenAI opens up all sorts of opportunities, whether it’s making the individual more impactful and giving them better tools and information, or tackling specific challenges such as how we help out trade customers when they’re on site,” says Sturdee.

For instance, if a tradesperson has a question around fitting a product or requires some technical knowledge, new technology powered by GenAI offers a way to have these questions answered.

This is essentially the use case for Chip. “Built using Microsoft AI technology, the tool gives our trade account holders the option to access product and fitting support at the touch of a button, meaning we’re there for our trade customers wherever and whenever they need us,” he says. “Chip marks only the start of Howdens’ AI journey, and we will continue to explore other ways AI can improve efficiencies for our customers.”

For Sturdee, AI models have moved on significantly in recent years. “They’re multimodal now,” he says, meaning they are able to handle different types of data, not just text input. “These new AI models are capable of taking vast amounts of information and making them accessible, and surfacing that information in a very meaningful way.”

AI in kitchen design

A major part of Howdens’ business is kitchen design. Sturdee sees an opportunity in the design process to use GenAI tools alongside designers, with the AI working behind the scenes across the design process.

“You have your general design, but then you’ve got all the pieces such as the skirting and the legs and the other pieces that go into a complete kitchen design,” he says. “An AI model removes a lot of that kind of drudgery work.”

Thanks to the power of deep learning, Sturdee also sees potential for GenAI to create entirely new things. With AI, he believes Howdens’ kitchen designers could have a rich user experience that streamlines the design process.

“You give them access to trends and perhaps even generate the first iteration of the design using an AI model,” he says. “Then the designers come in and use their expertise to finalise it.”

Sturdee believes such a design process is something that will be entirely achievable in the not-too-distant future.

Olaf Akkerman, general manager for retail and consumer at Microsoft, says: “If you think about that design stage, it’s quite an emotional buy for people, and you therefore go through many iterations and look for expertise with the tradespeople, as well as with Howdens, to get the best advice.”

Once on-site, the kitchen fitter may find the design needs adapting, and AI not only improves design efficiency, he says, but also provides a connection between Howdens, the tradespeople and the consumer. 

For the kitchen design process, Howdens is currently working with Microsoft on a feasibility study looking at whether the data structures are correct. “You need to make sure that the outcome is spot on,” says Akkerman. 

Looking at the evolution of the company’s first foray into the world of GenAI, Howdens initially worked with Microsoft to take in ideas from people across the business on where AI could be applied, and then assessed the business impact. “We came back with a shortlist,” says Akkerman. “Then we ran some quick experiments for feasibility. This resulted in Chip as the first use case.”

While a lot of AI use cases are internally focused, he says the way the team at Howdens has deployed AI in Chip shows the use of the technology as a digital assistant that sits on top of a knowledge base.

From the first conversation on AI, the whole process that resulted in Chip being built and deployed in the Howdens app took between six to nine months.

But, as Sturdee notes, the project will continue to evolve. “We will also continue to collaborate with Microsoft on the evolution of Chip, as it continues to evolve to meet the demands of our industry and trade customers,” he says.