Microsoft Frontier Tour London: Building trust in the agentic AI era

At Microsoft’s Frontier Tour London attended by 6,500 people, leaders argued that the next phase of artificial intelligence will depend on more than capable technology. Trust, human control, security and skills will determine whether organisations turn AI adoption into benefits people can feel. 

Artificial intelligence is already helping UK workers save time, find information, and tackle complex tasks. But as AI systems become capable of taking more actions on people’s behalf, a harder question is moving to the centre of the debate: how do organisations make sure people can trust them? 

That question ran through Microsoft’s Frontier Tour in London on 21 October, where Microsoft Chairman and CEO Satya Nadella and UK and Ireland CEO Darren Hardman set out what they see as the next phase of AI at work. Their message was that progress will depend not simply on adopting more capable technology, but on giving people more control, protecting organisational information, and building confidence in how AI is used. 

Trust by design 

Nadella framed the challenge around three needs: helping individuals retain agency; helping organisations make better use of their knowledge; and establishing trust in increasingly capable AI agents. These agents are AI assistants designed to carry out specific tasks or multi-step workflows with human oversight. 

“You’re feeling that sense of empowerment. You still have agency. You have control,” Nadella said to a packed ExCel centre audience, as he described Copilot as part of what Microsoft sees as a new operating system for work. 

For organisations, that also means being deliberate about what AI can access and do. Microsoft IQ brings together organisational context from sources such as documents, databases and messages so AI can work with relevant information. The principle is straightforward: useful AI needs context, but access to that context has to respect the organisation’s controls. 

Security becomes more important again when an AI system can act rather than simply answer. Nadella said organisations should treat AI models as “new forms of insiders” and build a zero-trust architecture around them. Microsoft’s Project Perception uses specialised AI agents to test safeguards, protect against incidents and help mitigate problems, while people continue to set strategy and retain judgement over important actions.

Nadella also featured several organisations in his keynote – easyJet, Unilever, Rolls-Royce; 1st Central and the UK’s National Health Service – that he thought were displaying innovative ways of applying AI to their operations.

Proof in practice 

Hardman connected that focus on trust with what is already happening across the UK. Microsoft research cited at the event ranks the UK eighth among 147 economies for generative AI adoption. Four-in-five business leaders believe the UK can become a world-leading AI economy within five years, but only two-in-five employees share that confidence. 

“We need to close that gap,” Hardman said. Trust is not built by ambition alone. It is built when people can see where AI helps, understand how it is governed and feel confident using it. 

He gave several practical examples. Network Rail is developing AI agents through Microsoft Copilot Studio for tasks including managing information about its 10,000-vehicle road fleet and supporting health and safety work. The AA is using AI to help 2,700 roadside patrol engineers find information across 18,000 technical documents. UK Power Networks uses Azure AI to help manage incoming emails and safety alerts, while AI agents help colleagues navigate complex documentation. 

And in public services, Hardman pointed to a trial of Microsoft Copilot across 90 NHS organisations, where users reported average savings of 43 minutes a day on administrative tasks. As a result of this success, NHS England is now rolling out Copilot to more than half a million clinicians and support staff to help relieve the admin burden and free up more time for patient care.

Dame Caroline Clarke, CEO, Online NHS Trust, joined Hardman on stage to discuss the ways the NHS is using AI, while emphasising the need for trust in the technology.

BT is also introducing Copilot through its ‘AI for All’ initiative, with employees reporting time savings of up to two hours a day. Other UK businesses are applying AI to challenges: Wordsmith is applying AI to legal work; Origin Tech to water leak detection; Red Flag Alert to financial risk; Altium to helping fraud investigations; and Wayve to autonomous driving, as Hardman discussed on-stage with Wayve CEO Alex Kendall.

Confidence to act 

For Hardman, the test is whether adoption changes the way work is actually done and produces benefits that people notice. He described that ambition as “bending the curve”: accelerating the pace of innovation and intentionally changing what the UK can achieve through higher productivity, business growth, and better public services. 

But the event’s clearest lesson was that capability and confidence have to develop together. Hardman said 71% of employees using AI report a positive impact on their work and career, while nine-in-10 business leaders report having a clear AI strategy. BUt Strategy alone, however, is not enough if workers do not understand the technology, trust its use or have the skills to apply it well. 

That puts people back at the centre of the story. Skills, security, governance and human judgement are not barriers to AI adoption; they are essential to it. 

As Hardman put it: “We have the talent. We have the ambition. Now we must give more people the opportunity to turn both into progress.”

The thousands of attendees were able to get hands-on with AI technologies, through breakout sessions, interactive learning opportunities, and plenty of networking.

For organisations considering what comes next, the practical challenge is to build that opportunity on trust: clear controls, human oversight, useful skills, and evidence of where AI genuinely improves work.