A new McKinsey report outlines a paradox in enterprise AI adoption. Around 88% of organisations now apply AI in at least one function, up from 78% the previous year, yet only a third have managed to scale it beyond pilots. The question: why the slowdown, and what separates companies seeing impact from those still experimenting?
Autonomous AI agents capable of planning and executing multistep tasks remain the biggest headline trend. Some 62% of respondents report experimenting with agents, and 23% say deployments are already scaling. Still, marketers overpromise: most enterprises remain in trial phases and moving to scale requires both deeper engineering work and process redesign. Agents demonstrate the most traction in IT and knowledge management, particularly in the tech, telecom, and healthcare sectors.
Financial performance remains muted. Only 39% report any EBIT impact, and for most the gain stays below 5%. Qualitative measures tell a different story: 64% credit AI with driving innovation, and roughly half say customer satisfaction and competitive positioning have improved. But adoption outpaces actual transformation. Many companies deploy tools without rebuilding workflows, infrastructure or guardrails. Scale is complicated for smaller players, and organisations with over 5 billion dollars in revenue scale nearly twice as often.
Around 6% of respondents report a significant business impact from AI. These leaders share a few traits. They pursue full business transformation rather than incremental cost-cutting. They invest heavily, with more than a third allocating over 20% of digital budgets to AI. Senior executives also stay hands-on: leaders are three times more likely to say management does not just voice support but actively adopts AI, participates in development, and ensures funding.
Despite the hype, success hinges on people. Leading organisations clearly define where humans interact with algorithms, when to validate output, and when expert judgement matters. AI rarely works in isolation. The greatest value lies in “hybrid intelligence,” which combines models with human experience. For conversational platforms, including players like Tovie AI, the signal is clear: growth will come from workflow-integrated systems and agentic solutions able to run complex dialogues and execute tasks autonomously. Winners will balance automation with oversight.
The gap between use and impact
The divide between adoption and business results shows that technology alone does not transform companies. Infrastructure, processes, culture and human trust matter just as much. In 2025, the businesses earning real money from AI are not necessarily those with the most advanced models; they are the ones that have integrated them properly. As the year continues, the question shifts from whether to adopt AI to how to scale it. Strategy, investment and a willingness to rethink work are now the deciding variables. For most organisations, the journey is just beginning.