As someone who’s watched 'AI transformation' turn into a stack of pilots and a few shiny demos, I understand why many CX and operations leaders feel conflicted right now.
The technology is moving fast, executives expect visible wins, and customers have zero patience for half-working automation.
McKinsey’s The State of AI in 2025 puts real numbers behind that tension.
On the one hand, AI is now mainstream. McKinsey reports that "88% of survey respondents say their organizations regularly use AI in at least one business function," and "72% report using gen AI (up from 33% in 2024)." On the other hand, the scale story is still shaky. The report finds that "nearly two-thirds have not yet begun scaling AI across the enterprise."
For CX and ops leaders, that gap is the point. Customers are increasingly moving through AI-mediated journeys, but many enterprises still lack the operating discipline to make those journeys reliable.
AI Has Moved Beyond Hype, and That Changes The Standard
One of the strongest positive takeaways from the analyst community is that the conversation has changed from experimentation to business transformation.
Gagan Bhasin, Founder & CEO VAO, sums it up in a way that mirrors the report’s underlying message about 'high performers':
"AI is no longer a futuristic concept, it’s a critical driver of business transformation."
That framing matters for CX leaders because 'transformation' in service is not a slogan. It means rethinking resolution, redesigning workflows, and deciding what humans should do versus what machines can safely do.
McKinsey’s own segmentation supports this. It defines a small group of 'AI high performers,' roughly 6% of respondents, as organizations that report significant value and attribute more than 5% of EBIT to AI.
That small number is both encouraging and sobering. It suggests real value is possible, but it is rare.
The Numbers Show Maturity, But They Also Reveal Fragility
If you want a single metric that captures 2025’s shift, it is still that adoption figure, 88% of organizations now use AI.
That market maturity creates a new problem for CX teams. When adoption is nearly universal, 'we’re using AI' stops being a differentiator. Customers won’t reward you for having a bot. They will reward you for reducing effort, increasing clarity, and fixing issues on the first try.
McKinsey’s data suggests AI is already moving needles in areas CX leaders care about.
In Exhibit 6, the report lists that AI can improve customer satisfaction by 45%. But the report also shows how uneven progress remains. Only 39% report any EBIT impact attributable to AI. That implies many organizations are still struggling to tie AI work to durable business outcomes.
In practice, CX leaders see the same pattern: visible activity, limited operational change.
The Report’s Biggest Future Signal: Agents That Do Work, Not Just Talk
The other major positive signal is the shift from chatbots to agentic systems.
Abhijit Verekar, Founder and CEO of Avèro Advisors summarizing the report in a LinkedIn post, frames the shift clearly:
“Agents are the new frontier: We are moving from chatbots that talk to agents that do.”
This aligns with McKinsey’s own data. The report finds 62% are at least experimenting with agents, and 23% are scaling agentic systems somewhere. In CX terms, that’s a directional move toward systems that can complete multi-step workflows, not just generate text.
But agentic AI also raises the bar for service operations. When AI can act, errors can become actions. So the core leadership question shifts from 'Can it answer?' to 'Can it execute safely, consistently, and with accountability?'
McKinsey’s risk findings suggest this is not theoretical. The report says 51% experienced at least one negative consequence, and that inaccuracy is the most common, with 30% experienced.
For service leaders, that’s an uncomfortable baseline. Inaccuracy is not a rare edge case. It is a normal operating risk.
The Critical View: Many Organizations Automated Tasks, Not Operations
Now to the harder part. Several critics argue that McKinsey’s data accidentally reveals how shallow many AI deployments still are.
Brianna Bentler, Co-Founder and CEO of Stealth AI captures this critique bluntly in a LinkedIn post:
"Zero integration. They’d automated tasks, not transformed operations."
Even without adding new data, this critique maps onto what McKinsey reports about scaling.
If nearly two-thirds of organizations have not begun scaling across the enterprise, then many deployments are likely isolated to teams, tools, or point use cases. That is where automation lives, and where transformation struggles to show up.