The State of AI 2025: How Leading Companies Are Turning Potential into Measurable Value
AI adoption keeps growing, and McKinsey's 2025 global survey shows which companies get the value: those that redesign their workflows around AI and put senior leaders in charge of it. Only 21% of companies using generative AI have fundamentally redesigned at least some of their workflows (McKinsey, The state of AI: How organizations are rewiring to capture value, 2025). The survey covers organisations of all sizes across industries and regions. This article summarises five findings and what they mean in practice.
Five findings
- Leadership drives results. Companies where the CEO oversees AI governance report the strongest effect on profit. When AI is left to IT alone, adoption moves more slowly.
- Workflow redesign brings the gains. Building AI into core processes creates the measurable improvements, and it is the step most companies still have ahead of them.
- Governance pays off. Leading companies manage risk and data centrally. Their top concerns are inaccuracy, cybersecurity and intellectual property.
- Scaling needs structure. Roadmaps, KPI tracking and role-based training help adoption grow across teams.
- Financial impact is appearing. More companies report higher revenue and lower costs in the units that use AI, especially in marketing, product development and service operations. These gains appear where AI is part of the process itself.
What this means in practice
- Give AI governance a senior owner.
- Redesign each process so AI carries out the work inside it.
- Set risk and compliance rules early, including the EU AI Act and GDPR.
- Define KPIs before the build and track them after launch.
- Train people for their new tasks.
- Start with one process and use what you learn for the next.
Why so few companies redesign
Redesign takes more effort than buying a tool. It means mapping how the work runs today, deciding which steps the system takes over, connecting it to your CRM and ERP and changing people's roles. Companies that do it once build the skills to do it again, and each new process goes faster.
What redesign looks like
In our Invoice Processing project, a manufacturer redesigned invoice handling around an AI agent. The agent reads each invoice, matches it to the order and posts it to the ERP, and the team handles the exceptions. It took over the work of three positions, and seasonal hiring ended. The project went live in weeks, at a fixed price agreed before the build.
Where to start
Pick one process, redesign it from start to finish and measure the result. If you want the full picture first, AI Strategy Consulting gives you a roadmap with the cost and payback of every task.