The promise of agentic artificial intelligence — AI systems capable of autonomously making decisions and executing complex tasks — has excited enterprises and tech giants worldwide. Yet a recent Gartner report forecasts a sobering reality: more than 40% of current agentic AI projects will be cancelled by 2027. This anticipated shakeup reflects rising costs, unclear business value, and implementation challenges that threaten many initiatives before they mature. 

The hype and the reality: Gartner’s forecast

Gartner’s analysis highlights a key tension in the agentic AI market today. While companies like Salesforce, Oracle, Microsoft, and others have poured billions into AI agents aiming to boost efficiency and automate operations, many projects are still in nascent phases — largely experiments or proofs of concept. Gartner’s senior director analyst Anushree Verma warns that much of the industry hype around agentic AI is currently overshadowing its true capabilities.

A major issue is “agent washing”, where vendors rebrand traditional chatbots, AI assistants, or robotic process automation tools as agentic AI without the systems demonstrating genuine autonomy or intelligence. Gartner estimates there are only about 130 vendors globally with truly agentic AI capabilities among thousands claiming the label. This gap between marketing claims and technological reality is a key driver behind the predicted project cancellations.

However, Gartner projects significant growth beyond the short term. By 2028, an estimated 15% of routine daily work decisions will be autonomously handled by AI agents, up from near zero today. Additionally, 33% of enterprise software applications will incorporate agentic AI, compared to less than 1% now. This suggests a market still in flux — many projects will fail, but those that succeed will unlock transformative value.

Agentic AI as a game-changer in customer service?

Despite the setbacks predicted, agentic AI is poised to revolutionise certain sectors, with customer service leading the way. Gartner’s March 2025 report predicted that by 2029, 80% of common customer service issues would be resolved autonomously by AI agents, leading to a 30% reduction in operational costs. This evolution moves beyond traditional AI chatbots, which could only answer questions or summarise interactions, to AI agents that take independent action — like cancelling subscriptions, negotiating rates, or proactively fixing issues before customers even notice.

Daniel O’Sullivan, senior director analyst at Gartner, calls agentic AI a “game-changer” for customer service. He highlights that organisations will need to:

• Upgrade infrastructure and optimise self-service tools to handle increased AI interactions,
• Adapt service models to differentiate between human and AI-generated requests,
• Establish policies for AI-led interactions ensuring data privacy, security, and appropriate escalation, and
• Collaborate closely with product teams to integrate AI proactively.

This shift would fundamentally reshape the customer experience and how service teams operate, ushering in a new era of efficient, low-effort service delivery driven by AI automation.

Agentic AI as enterprise brain: Insights from the World Economic Forum

Meanwhile, a World Economic Forum report adds another dimension, portraying agentic AI not just as isolated tools but as the “brain” of the enterprise. Current enterprise AI agent deployments often remain fragmented, limited to specific departments, and lacking sophistication. True transformation demands connecting specialised agents into a cohesive, strategy-aligned system spanning functions, such as finance, operations, marketing, and customer service.

The report emphasises the need for specialised enterprise AI platforms — analogous to the Oracles and SAPs of previous tech eras — that can deploy, manage, and evolve AI agents at scale. Companies must progress through a maturity journey:

1. Breadth: Extending agent use across more departments,
2. Depth: Enhancing the sophistication and autonomy of agents, and
3. Integration: Ensuring seamless coordination among agents.

Achieving this will enable companies to become “cognitive organisations” — intelligent, adaptive, and aligned end-to-end. Looking forward, CEOs would manage hybrid workforces combining human employees and AI agents. Some experts even envision a future where entire companies operate autonomously, managed by networks of AI agents with minimal human involvement.

Challenges: KPMG’s perspective on implementation

While optimism about agentic AI’s potential remains strong, KPMG’s research tempers expectations by underscoring real-world hurdles. Technical skill shortages, employee resistance to change, and complex system integration create significant barriers to deploying agentic AI at scale. These challenges slow progress and contribute to Gartner’s prediction of widespread project cancellations.

KPMG’s survey reveals that nearly half (46%) of business leaders focus their agentic AI strategies on improving efficiency and revenue growth. However, concerns about data privacy, regulatory compliance, and data quality are at a three-quarter high, highlighting the need for a thoughtful, strategic approach rather than rushing headlong into AI deployment.

Notably, major players, such as Microsoft, Salesforce, Oracle, SAP, and Workday are actively launching agentic AI products, while consulting firms KPMG and Deloitte offer their solutions. This signals an industry-wide commitment to overcome early hurdles and capitalise on AI’s transformative promise.

The future of work: BCG on AI agents as teammates

The Boston Consulting Group (BCG) provides a forward-looking view on how AI agents will reshape the workforce. With the agentic AI market expected to grow at a 45% compound annual growth rate (CAGR) over the next five years, AI agents will increasingly function as teammates alongside humans.

BCG predicts that complex disciplines — including software development, customer service, and business analytics — will shift from large human teams to smaller groups working with AI agents. This will enable organisations to scale faster, as AI agents can be replicated quickly without the constraints of traditional hiring.

AI agents will unlock new business models, boost productivity, and free human workers from repetitive tasks to focus on creativity and strategy. However, managing these AI agents will become a core skill for employees, requiring training in responsible AI use to uphold privacy, ethics, and fairness.

Synthesis: Navigating the agentic AI landscape

The collective insights from Gartner, the World Economic Forum, KPMG, and BCG paint a nuanced picture of agentic AI’s trajectory. On one hand, Gartner’s forecast that over 40% of projects will be scrapped by 2027 reveals the industry’s current overenthusiasm, technical challenges, and vendor marketing issues. Many initiatives today are experiments lacking clear value propositions, and “agent washing” risks eroding trust in genuine AI progress.

While, multiple reports affirm that agentic AI holds transformative potential — particularly in customer service, enterprise-wide coordination, and workforce augmentation. The technology is progressing steadily toward making autonomous work decisions and embedding itself deeply into enterprise software ecosystems.

Yet realisation of this potential requires overcoming significant barriers: skills shortages, cultural resistance, integration complexity, and heightened concerns about data privacy and ethics. Its maturity, as well as enterprises’ sophistication in deploying it, will be critical to understand whether agentic AI projects succeed or fail.

Moreover, agentic AI is still in its early innings. The coming years will be a period of shakeout — with many projects discontinued — but also a time of consolidation and innovation. Enterprises that adopt a strategic, measured approach, investing in infrastructure, governance, workforce training, and cross-functional coordination, will be best positioned to harness its power.