The biggest mistake leaders can make with AI is to treat it like another software rollout.
That was one of the clearest messages from Bosch Connected World 2026 in Berlin, where Katy George, Corporate Vice President of Workforce Transformation at Microsoft, gave a fascinating inside view of how Microsoft is preparing its own workforce for the AI era.
Microsoft has a unique vantage point. As George put it, the company is “customer zero” for its own products and solutions. It is using AI internally, studying what works and codifying lessons for others. That makes its experience especially useful for any business moving from AI experiments to real transformation.
George’s core argument was refreshingly practical. AI transformation is not about sprinkling copilots and agents across old processes. It is about understanding how work actually happens, where value is created and where AI can change the operating model of the business.
How Microsoft Is Redesigning Work
Microsoft began by studying 100 internal AI transformation case studies. From that work, George’s team identified three patterns for using AI to improve business performance.
The first is the “Persona Accelerator.” This means taking a role where many people do similar work, studying the daily tasks in detail and identifying the prompts, copilots or agents that can help people perform better.
The second is end-to-end process redesign. Microsoft uses methods familiar from lean and continuous improvement, such as Gemba walks and value stream maps, then applies them to knowledge work. That is harder than it sounds, because knowledge work is often invisible, inconsistent and buried in emails, meetings and habits.
The third pattern starts from scratch. Instead of improving an existing process, teams design an AI-first way of working around the desired input and output.
This is where the bigger opportunity sits. AI is useful when it removes friction, but far more powerful when it allows businesses to create new value.
George made that point clearly. Productivity is a benefit, but it is not always the main prize. As she said, “The business results, the strategic results that we’re able to deliver for our business actually go way beyond labor cost replacement.”
That is an important message for leaders. If AI is measured only by time saved or costs reduced, organizations may miss the bigger opportunity. AI can improve quality, increase speed, reduce risk and create services that were previously impossible.
George gave the example of Microsoft’s internal audit function. AI is helping audit teams work faster and cover more of the business, but the bigger shift is in what audit can now deliver. Instead of mainly reviewing what has already happened, audit can use AI to spot potential risks earlier and bring those insights into every engagement. That makes the function more proactive, more scalable and more valuable to the business.
Why AI Transformation Is A Business Transformation
One of George’s strongest lines should be required reading for every executive team: “We can’t treat this like a tech project, and it is not a product launch.”
AI transformation touches decisions, workflows, roles, skills, risk and leadership. It cannot be delegated to IT as a tool deployment. Business leaders need to define the goals, own the outcomes and engage directly with how work changes.
At Microsoft, the aim is broader than productivity alone. George talked about AI improving revenue, quality and speed. Sales is a good example. Microsoft uses AI to help sellers prepare for customer conversations, rehearse different scenarios with a coaching agent and navigate complex deals. It is also using AI to reach smaller customers who may not have been served directly by human sales teams before. The result is a sales function that can be more effective, more personalized and more scalable.
The question is not simply which tasks can be automated. The better question is how AI changes the relationship between people, process and performance.
Making Invisible Work Visible
To scale AI, organizations first need to make work visible. In factories, you can often observe the process. In offices, the process may be hidden inside messages, spreadsheets, meetings and informal workarounds. People may know how things get done, but the organization may struggle to describe it.
George called knowledge work “tacit, invisible, non-standard.” That is why simply adding AI tools is not enough. Leaders need to understand workflows, handoffs, decision rights, data flows and quality standards before they can redesign work effectively.
The people closest to the work are central to this. As George put it, “Only the people who know the work can actually reinvent the work.”
That line captures one of the most important lessons of AI transformation. Employees are not passengers in this shift. They are the people who will discover, test and improve the new ways of working.
The Human Side Of AI
George was also refreshingly honest about the human impact. She acknowledged that AI is “nerve-wracking” for employees, even at Microsoft. If people inside one of the world’s leading AI companies feel uncertain, leaders everywhere should assume their own teams do too.
Microsoft’s response is to focus on careers rather than job permanence. Every job will change. The commitment is to help people learn, adapt and remain valuable in an AI-shaped economy.
That is a better message than pretending nothing will change. Employees can see what AI can do. Trust comes from honesty, support and a credible path forward.
George also shared an interesting internal signal. Microsoft tracks how AI use affects employee engagement, through what it calls Thrive scores. She said, “Our employees who use AI the most are also our happiest.”
That does not mean AI automatically improves work. Poorly deployed AI can create confusion, pressure and noise. But when it removes repetitive toil and helps people do more meaningful work, it can improve both performance and engagement.
Leadership Sets The Pace
George ended with Microsoft’s biggest lesson: “It is really all about leadership.”
Many leaders want AI-powered organizations while continuing to work in old ways themselves. They approve budgets, encourage experimentation and talk about transformation, yet their own behavior barely changes.
George was direct: “Those leaders who are power users themselves have organizations that are power users. There is no substitute for personal role modeling.”
That is the leadership challenge in one sentence. Executives need to show how AI is changing their own work, in meetings, decisions, customer preparation, research and communication. They need to be visible learners.
They also need to set risk boundaries. George stressed the importance of deciding up front where humans must remain in control and where AI can have greater autonomy. As agents become more capable, clarity around accountability becomes essential.
The Future Of Work Is Still A Human Choice
Near the end of her keynote, George made one of the most important points of the day: “nothing is technologically preordained.”
AI will reshape work, but the outcome is not fixed. It depends on the choices leaders make, the boundaries they set, the skills they build and the values they embed into their organizations.
Microsoft may be one of the companies building the AI future, but its own experience shows that the human choices around leadership, trust and work design will matter just as much as the tools.
