By Kasturi Datta |
Microsoft chairman and CEO Satya Nadella spoke about what it takes to become a ‘frontier’ organisation by exploring how businesses can move beyond experimenting with AI to redesign processes around AI agents, as well as highlighting the latest real-world AI use cases from the UK at Microsoft’s AI Tour London.
Nadella began the keynote with a focus on how businesses should refine their approaches to integrating AI into their workflows. “Grounding ourselves in ensuring that AI translates into real world impact is the most important thing we can do,” he said.
Microsoft’s AI stack is designed for frontier transformation, operating as a three-layered system that is founded on the relationship between the “intelligence layer” and trust in the governance of AI models.
The top layer of the stack is high-value agentic experiences, which involves integrating AI into existing workflows. “The thing we have gotten very used to is invoking agents, so we are able to hand tasks over to agents synchronously and have the agents work on our behalf,” said Nadella.
Nadella was joined by Anne Krupke, senior product marketing manager at Microsoft, to demonstrate agents in Microsoft 365 Copilot for finance operations. Krupke showed how to generate actionable insights and access relevant policy documentation to resolve financial discrepancies. Nadella summarised the demonstration as “macro-delegation with micro-steering”.
Manchester University NHS Foundation Trust is using Dragon Copilot to improve workflows and reduce cognitive burden for employees. The platform is being used to record patient appointments to automatically writ up electronic reports, giving clinicians more time to focus on connecting with their patients. “We are trying to reduce the number of clicks for our clinicians, and this is exactly what we are achieving with Dragon Copilot," said Ceyda Mogulkoc-Zhuang, director of the electronic patient record and digital applications at Manchester University NHS Foundation Trust.
NHS doctors are using Dragon Copilot to better connect with patients
The second layer of the AI stack is agent platform, which creates high value agentic experiences like tools in Dragon Copilot. Sitting between the first and second layer is the need for unified intelligence, comprised of Work IQ, Fabric IQ and Foundry IQ.
Microsoft Foundry’s core competency is the delivery of models to build and govern AI apps and agents. The platform now has over 11,000 models, including the recently added GigaTIME, which uses generative AI to translate the contents of pathology slides into a simulation that can be used to chart immunotherapy progression for patients, a process that traditionally would have required a lot of time and money.
“Our hope is to empower every secondary hospital in every tertiary city to be able to conduct their own research,” said Hoifung Poon, general manager of Microsoft Research.
The final, bottom layer of the AI stack is the token factory, which powers the agentic AI tools. “We’re ensuring that the token factory delivers on the core formula, which is tokens per pound per watt,” said Nadella.
There are currently over 400 Microsoft data centres and 70 Azure regions around the world. Maia 200 is Microsoft’s custom-built AI accelerator chip designed to efficiently run large-scale AI interference across these data centres.
Nadella then spoke about Microsoft’s commitment to data sovereignty: “We have to ensure that we have sovereign controls for every region and country.”
Updates to Microsoft Sovereign Cloud – including Azure Local disconnected operations, Microsoft 365 Local disconnected and Foundry Local – let organisations run critical infrastructure and productivity apps with Azure governance and policy controls, without the need for cloud connections or inside their on-premises, private cloud environments.
Darren Hardman, CEO of Microsoft UK and Ireland, then highlighted some of the most recent Microsoft AI use cases.
“At Microsoft, we believe that AI has the power to unlock innovation, phase out the mundane and unleash human creativity on a scale never seen before,” he said.
Microsoft has committed $30 billion dollars to expand AI and cloud infrastructure across the UK. Microsoft will also work with Nscale to build the country’s largest supercomputer with more than 23,000 NVIDIA graphics processing units.
Hardman reinforced Nadella’s belief that to maximise the full potential of AI, organisations must commit to becoming frontier and build on their intelligence layer, saying that “frontier transformation has the potential to drive growth across every industry.”
Construction giant Balfour Beatty has reported 75 per cent of its workforce feel that Microsoft Copilot has improved the quality of their work and 77 per cent have reported less mental effort. Microsoft has also been working in close partnership with the UK Government and is a founding partner of its AI Skills Boost, a programme with the aim to skill 10 million workers in AI competency by 2030.
“For more than a decade, Microsoft’s mission has been to empower every person and every organisation to achieve more,” said Hardman. “And what excites me so deeply about this moment and about becoming frontier, is that AI allows us to live that mission more fully than ever before.”