When presented with so many options, avenues, and potential outcomes for AI tools, some companies will inevitably struggle to build a strategy. That’s why learning from other companies’ experiences is so important and useful — and there’s perhaps no better company to learn from than the one making the AI.
To that end, Microsoft has shared what it has learned from using AI internally. It has conveniently outlined these lessons in five key findings that I’ll present along with other examples I’ve come across to reinforce these ideas.
Five Key Lessons
“We have been sharing Microsoft’s Frontier Playbook with customers as a practical guide to our AI transformation journey, including what we’ve learned, what has worked so far and where we’ve grown from failures,” said Kathleen Hogan, Microsoft’s Executive Vice President and Chief Strategy and Transformation Officer.
“Drawing on hundreds of AI transformation efforts across the company, the playbook captures what we are learning as we redesign work, build new capabilities, measure impact and help people grow alongside AI.
“Among the many insights gained from our successes and failures, five lessons consistently stand out.”
Start with the business outcome, not the technology: Hogan explains that simply giving employees AI tools doesn’t guarantee transformation. So, to address this, Microsoft shifted its focus from AI adoption to business goals, things like winning deals and delivering better customer value.
Looking at early usage of the technology in sales, Microsoft identified specific points where AI could help account managers the most and deployed agents to tackle those use cases. As a result, priority AI use-case adoption tripled, revenue per account manager increased 9.4%, and close rates were 20% higher.
This is a point Microsoft has been pushing for a while now. In fact, back in January 2025, Microsoft’s Charles Lamanna, Corporate Vice President, Business & Industry Copilot, made a similar point to Cloud Wars founder Bob Evans: “You already know the business outcomes that matter. There’s already metrics that you track. Just go to those metrics and apply AI there.”
Redesign the entire workflow, not just individual tasks: Microsoft has recognized that adding AI to broken processes doesn’t fix the underlying issues. As Hogan puts it, “speeding up one step just creates a longer queue at the next.” Instead, the best results came when teams focused on the entire workflow, as was the case with Microsoft’s cloud supply chain team.
The team first simplified its workflows and created a single source of truth for data that the agentic system could collectively tap into. Next, it deployed more than 100 AI agents across planning, sourcing, fulfilment, and logistics.
Some workflows saw cycle times fall by up to 75%, while a number of investigations went from taking 5–7 days to less than 20 minutes.
“We’ve found the largest gains come when teams step back and redesign how work should flow across people, process and technology from start to finish,” said Hogan.
Put employees at the center of transformation: “The people who do the work know where processes break down, where judgment matters and where AI could help — insights that no process map can fully capture,” says Hogan.
Ultimately, employees understand where the processes they use break down, where AI can help, and where human judgment is needed. And this is a similar conclusion to the one Workday has reached. I covered the research behind that thinking in my article, Beyond Productivity: How AI + People Drives Lasting Business Value. Workday’s research also points to the importance of investing in people, skills and changes to roles and processes alongside AI deployment.
Microsoft’s Camp AIR program helps teams experiment with AI and redesign how they work around real business challenges. To date, the program has scaled to more than 3,000 engineers across the organization.
One team to take advantage of this was the Copilot Cowork team, a nine-person group of engineers, designers and product managers that used the multi-week AI transformation accelerator to rethink how a product gets built, with AI embedded from day one. Working alongside AI agents, the team eventually shipped an initial release in just 35 days.
Use AI to expand what people can do: While Microsoft’s initial focus was on automation and efficiency, the company now sees a bigger opportunity in what it calls “Capability Add” — combining human and AI capabilities to achieve things that were previously impractical or even impossible.
“Think of it as an equation: CI + AI = CA. Continuous improvement takes waste out and AI adds capability in,” says Hogan. “Together, they produce Capability Add — output with higher strategic value.”
Within the wider Cloud Wars ecosystem, during a roundtable in 2025, ServiceNow’s Dorit Zilbershot explained a stance that echoes Microsoft’s thinking:
“All of us are really becoming a manager in the future. This means that the AI agents will be doing all the routine, repetitive, time-intensive tasks, while we as humans really focus on high-level strategy, creativity, problem-solving and that collaboration between agents under that AI orchestration.”
Combine human and AI capability to create a continuously learning organization: Microsoft has found that AI can help employees learn, experiment and take on more complex work. From the employee side, people can give AI context, feedback, judgment, and knowledge.
This, explains Hogan, creates a continuous learning loop, where people improve AI while AI helps people improve. “Over time,” she says, “this learning loop turns individual insights into organizational capability.”
Hogan specifically describes AI as a “thought partner” — and I have to admit, it’s the first time I’ve heard that term. I’ll probably be repurposing it in the future, because it neatly captures many of the ideas floating around about how we should think about and position AI: not simply as a tool that executes tasks, but as something we can work with, learn from, and use to develop new ideas.
