Where & how to use gen AI, the MS way

How do you build unique, industry-specific AI solutions that are practical, useful, and coveted by companies operating in vastly different domains? This was the question the Industry AI team at Microsoft asked themselves right after the world was exposed to the amazing capabilities of ChatGPT and generative AI systems.
The first step they took to find answers was to start talking to their network of partner companies in socalled ‘ideation workshops.’
“It was our way to share what we know about AI, what we’ve learned, and what we’ve experienced in terms of what’s working. But also, what we’ve seen as challenges; and it was a chance to learn from top industry partners on what they’re seeing and what it means for their world,” Merav Davidson, VP of Industry AI at Microsoft, told us on a visit to India.


During these workshops, Merav, and the rest of the Industry AI team, noticed that companies with advanced AI and natural language processing (NLP) capabilities found it easy to adopt and implement generative AI solutions into their work processes, but that the situation was vastly different in less tech centric companies.
“What we saw was a lot of excitement. The CEO would tell the employees they had to do something with this thing (Gen AI) that everyone was using, but nobody was sure what to do with it and how to do it,” says Merav.
After extensive talks during the workshop, Merav says her team would then come up with a list of use cases on how and where generative AI systems could be deployed for that customer. “We would end up with some prioritised use cases that would make a difference in their specific industry and would actually allow them to create a new experience, significantly improve an existing experience, differentiate, and maybe even to some extent disrupt the specific industries that they have been playing in.”


After talking to numerous companies over the past year, Merav says they now have a wealth of useful information related to how to best deploy Gen AI systems safely – taking into account hallucinations, data privacy worries, and responsible AI concerns. “We’ve collected a lot of great requirements and information in terms of what we can do to remove friction for companies who are adopting this technology. We’ve also started building components and tools and solutions to help close gaps and remove those friction points.”
Creating templates
One organisation that Microsoft’s Industry AI team has been helping out is the London Stock Exchange Group (LSEG). Large organisations in the financial sector generate a lot of data, and meetings related to such financial service activities are inherently complex and heavy with numbers and graphs. The solution the Industry AI team came up with, called Meeting Prep, automatically generates meeting preparation summaries combining insights from LSEG’s financial data with news, documents, emails, chats, and other relevant content from Microsoft Graph in Microsoft 365.
The team has worked with Siemens to introduce Industrial Copilot, an AI-powered assistant that will enable users to rapidly generate, optimise, and debug complex automation code, and significantly shorten simulation times. It has worked with Bay er on an agricultural Copilot, looking to add value through the ability to interact with agronomic data using natural language. It has worked with edtech company Anthology to create AI Design Assistant, enabling instructors to spend more time with students and less time creating courses and assessments.


Merav says an analysis of a customer service Copilot found that it enabled 14% more cases to be resolved in an hour. “We also have stats that show information workers are saving more than an hour a week using Copilot. Developers can code 55% faster with it too,” she says.
The Industry AI team has released multiple templates that different industries can easily follow to launch generative AI solutions in their respective organisations. More templates are planned to be released in the coming months.

Leave a Comment

Your email address will not be published. Required fields are marked *