From Generative AI Hype to Enterprise Disappointment

Over the past two years, enterprises have accelerated their investments in Generative AI. According to industry surveys, 78% of companies have deployed GenAI tools, spanning everything from marketing copy to customer support scripts. These initiatives often centred on enhancing productivity and automating content-heavy workflows.

However, the results have largely failed to meet expectations. Nearly 80% of organisations report no material impact on their bottom line, despite widespread adoption. This disconnect is now referred to by analysts as the Generative AI conundrum: widespread implementation without commensurate business transformation.

Why? The majority of GenAI deployments have typically focused on narrow, horizontal use cases — deploying single-task copilots or chat-driven support assistants. While these are easy to pilot and quick to showcase, they rarely address core enterprise workflows or unlock new sources of value.

The Shift: Agentic AI as a New Operating Model

Agentic AI represents the next evolution, moving from systems that generate outputs to those that independently plan, decide, and execute actions across complex workflows. Unlike traditional automation or even advanced GenAI, Agentic AI systems are built to:

• Coordinate multi-step processes that cross functional boundaries.
• Adapt dynamically based on real-time data and embedded feedback loops.
• Integrate directly into enterprise systems (ERP, CRM, industry-specific platforms) to drive end-to-end outcomes.

By 2028, Gartner projects that 33% of enterprise applications will embed Agentic AI, with autonomous agents handling up to 15% of daily operational decisions. Early deployments already demonstrate meaningful impact:

• A leading insurer reduced underwriting timelines from two weeks to three hours by deploying a network of 78 AI agents that autonomously reviewed documentation and executed policy approvals.
• In healthcare, Oracle Health’s clinical AI agents have cut physician documentation time by 41%, returning nearly an hour per day to patient care.
• Klarna has leveraged AI agents in customer support to resolve two-thirds of chats end-to-end, achieving both cost reductions and higher customer satisfaction.

Why Most Organisations Aren’t Prepared

While the promise of Agentic AI is substantial, it introduces complexities that most enterprises are not yet structured to handle. Unlike GenAI pilots that could often operate in functional silos with human-in-the-loop safeguards, Agentic AI requires a foundational shift across three critical dimensions:

1. Governance and Control:
   Autonomous decision-making demands robust frameworks for security, auditability, and ethical oversight. Agentic systems must have clearly defined escalation protocols, transparent reasoning paths, and compliance checks embedded by design.
2. Organisational Readiness:
   Effective deployment requires data environments that can support composable, event-driven architectures. Many organisations still operate on fragmented, legacy stacks that limit the ability of agents to access and process enterprise-wide data in real-time.
3. Talent and Leadership:
   Traditional roles centred on execution will give way to new positions such as agent trainers, escalation architects, and AI risk officers. Leadership teams will need to cultivate a culture where managing AI counterparts becomes a core managerial competency.

The Strategic Leadership Imperative

Agentic AI is not a technology add-on; it is an inflexion point that compels leadership teams to reconsider how work is organised and value is created.

Example of key questions for executives:

• What competitive advantage can Agentic AI unlock beyond efficiency gains?
  (For example, dynamic pricing models that adapt to supply chain shifts in real-time or underwriting that learns continuously from evolving risk profiles.)
• How do we build trust into autonomous processes?
  (This includes transparent decision logging, bias detection, and human override protocols.)
• Are we prepared to make investments beyond tooling — in organisational redesign, change management, and digital fluency at scale?

Notably, early movers have treated Agentic AI as a strategic lever rather than a series of functional experiments. They’ve invested in cross-functional teams, embedded governance councils, and designed workflows where human and AI roles are intentionally interlinked.

The Road Ahead

This report marks the first installment in our “12 Questions for Leaders on Agentic AI” series, crafted to help organizations navigate this pivotal shift with clarity and intent. Future editions will delve deeper into the four critical pillars that will shape success in the Agentic AI era:

• Strategic Alignment: How to embed Agentic AI into the very fabric of your business model and competitive strategy.
• Value Realisation: How to move beyond pilots to measurable outcomes that drive margin, speed, and new growth opportunities.
• Organisational Readiness: How to build the data, technology, and talent foundations required to scale autonomous decision-making responsibly.
• Governance & Trust: How to design oversight, transparency, and ethical safeguards into systems where agents act with real autonomy.

For forward-looking enterprises, the question is no longer if they will adopt Agentic AI, bot how systematically and responsibly they will integrate it, and whether their leadership is prepared to translate this next wave of intelligence into enduring advantage.