Why 78% of companies using generative AI see zero material impact, and how agentic AI is rewriting the rules of business transformation
The numbers don't lie, and they're telling a story that should make every CEO pause. McKinsey's latest research reveals a disconnect so profound it's reshaping how I think about AI strategy entirely: 78% of companies are using generative AI, yet 80% report zero material impact on earnings.
I've been watching this paradox unfold across our client base for months now. Boardrooms are filled with excitement about AI adoption. Massive investments in enterprise copilots. Pilot programs are proliferating across every department. Yet when we dig into the actual business outcomes, the results are frustratingly incremental.
This isn't a technology problem; it's a strategic one. We've been playing in the shallow end of the AI pool, focusing on horizontal tools that provide diffuse benefits while the truly transformative vertical use cases remain trapped in what I call "pilot purgatory."
But something fundamental has shifted. The breakthrough isn't coming from better language models or more sophisticated prompts. It's coming from an entirely new category of AI that's redefining what's possible: agentic AI.
The Fundamental Shift from Reactive to Proactive
The difference between traditional generative AI and agentic AI isn't just technical; it's philosophical. Where large language models wait passively for human prompts, AI agents combine autonomy, memory, planning, and deep system integration to become proactive virtual collaborators.
They don't just respond to requests; they anticipate needs, orchestrate complex workflows, and adapt to changing conditions in real time. This isn't about making human tasks easier; it's about reimagining what work itself looks like.
I've witnessed this transformation firsthand. A major financial institution we worked with achieved more than a 50% reduction in legacy system modernisation time using coordinated agent squads. Not through automation of individual tasks, but through intelligent orchestration of entire transformation workflows.
A research firm saw 60% productivity gains when it deployed autonomous agents for data quality analysis. These weren't simple automation scripts; they were intelligent systems that could identify data anomalies, trace root causes, and implement corrections across multiple systems simultaneously.
Credit risk processes that traditionally took weeks now complete in days, with 20 60% efficiency gains through agentic workflow redesign. The agents don't just process applications; they proactively gather additional data, cross-reference multiple sources, and even initiate follow-up actions based on risk assessments.
The Strategic Inflection Point
What I'm seeing represents more than incremental improvement; it's a fundamental shift in how businesses can operate. Agentic AI enables four capabilities that were previously impossible:
Parallel execution that collapses traditional cycle times. Instead of sequential handoffs between departments, agents can orchestrate multiple workstreams simultaneously, dramatically reducing time to value.
Real-time adaptability to changing conditions. When market dynamics shift or unexpected challenges arise, agents can recalibrate entire processes without human intervention, maintaining momentum where traditional workflows would stall.
Elastic scaling that flexes with demand. Peak periods that previously required surge staffing or resulted in bottlenecks can now be absorbed seamlessly by agent networks that scale up and down automatically.
Proactive orchestration across complex processes. Agents don't wait for problems to escalate; they identify patterns, predict issues, and initiate corrective actions before human operators even realise something needs attention.
But here's what keeps me up at night: the biggest challenge isn't technical, it's organisational. The companies that will win in the agentic era are those that fundamentally reimagine their operating models.
Reframing the Strategic Question
We need to stop asking "Where can we use AI?" and start asking "What would our core processes look like if agents ran 60% of them?"
This isn't about replacing humans; it's about elevating human work to its highest value. When agents handle routine orchestration, monitoring, and coordination, humans can focus on strategy, creativity, and complex problem solving that requires genuine insight and judgment.
At Keyrus, we've developed a framework that addresses the four critical enablers for agentic success:
People transformation goes beyond traditional upskilling. We're helping organisations develop entirely new collaboration models between humans and agents, redefining job roles around strategic oversight rather than tactical execution.
Governance frameworks that enable autonomy while preventing chaos. Agents need clear boundaries and decision-making authority, but they also need guardrails that prevent uncontrolled proliferation and ensure alignment with business objectives.
Architectural foundations that support what we call the "agentic AI mesh," a distributed network of specialised agents that can communicate, coordinate, and collaborate across traditional system boundaries.
Data productization that provides the fuel agents need to operate effectively. This means moving beyond traditional data warehouses to create real-time, contextualised data products that agents can consume and act upon autonomously.
The Transformation Imperative
The window for gradual adoption is closing. Agent native competitors are emerging across industries, and they're not constrained by legacy processes or organisational inertia. They're building their operations from the ground up around agentic capabilities.
Three actions I'm recommending to every leadership team:
First, conclude your pilots and realign AI priorities around strategic business processes. Stop experimenting with horizontal tools and start building vertical solutions that directly impact your core value streams.
Second, redesign your governance with cross-functional transformation squads that can move quickly and decisively. Agentic AI requires organisational agility that traditional hierarchies can't support.
Third, launch lighthouse projects while simultaneously building the foundational capabilities that will enable broader agentic deployment. Don't wait for perfect conditions, start learning while you're building.
The companies that master this transition won't just gain efficiency, they'll redefine their competitive position entirely. They'll operate with speed, scale, and sophistication that their competitors simply can't match.
Those who don't risk something far worse than missed opportunities. They risk accelerating decline as the market moves beyond their ability to adapt.
The Gen AI paradox taught us that adoption without transformation is meaningless. The agentic AI era demands that we go beyond tools and technologies to fundamentally reimagine how business gets done.
The question isn't whether this transformation will happen; it's whether you'll lead it or be left behind by it.
Bruno Dehouck is CEO of Keyrus UK and Iberia, a global consultancy specialising in innovative data, AI, and digital technology solutions. Connect with him on LinkedIn to continue the conversation about agentic AI transformation.
🔗 Read the full McKinsey report: "Seizing the Agentic AI Advantage"