Generative AI Loses Its Shine as Marketers Turn to Data and Agents
Here’s what you’ll learn when you read this article:
Why generative AI has entered the Trough of Disillusionment and what that means for business value.
How foundational technologies like AI-ready data, ModelOps and AI engineering are becoming essential for scaling AI.
What marketers and communicators should do to reset expectations, strengthen governance and build trust in AI.
Generative AI has entered a new phase in Gartner’s 2025 Hype Cycle for Artificial Intelligence. Once touted as the most transformative technology in decades, GenAI is now sliding into what Gartner calls the Trough of Disillusionment. The shift highlights a critical moment for marketers and communicators, who must recalibrate how they position AI inside their organizations.
The Hype Cycle shows that after a year of heavy spending, the enthusiasm surrounding GenAI is giving way to skepticism. Gartner reports that companies invested an average of $1.9 million in GenAI initiatives in 2024, yet fewer than 30 percent of AI leaders say their CEOs are satisfied with the returns. Many organizations entered projects with unclear goals or unrealistic expectations. More advanced teams struggled to find talent and build literacy among employees who lacked a full understanding of the technology.
Marketers face the challenge of balancing opportunity with realism. GenAI is powerful, but its business impact requires the right use cases, clean and structured data, and an informed workforce. Communicators need to help executives, boards and employees understand that value comes through integration and governance, not one-off experiments or viral outputs.
Gartner’s analysis points to a shift in focus from the flashiest applications of AI to the enabling technologies that make it scalable. AI engineering and ModelOps, which govern the life cycle of models and analytics, are moving into the spotlight. These disciplines provide the structure for deploying AI consistently and securely at scale. For marketing teams, that translates into reliable personalization, campaign performance tracking and predictive insights that can stand up to scrutiny from both the C-suite and regulators.
Two other areas show strong movement in this year’s Hype Cycle. AI-ready data and AI agents are both at the Peak of Inflated Expectations. Data readiness is now a pressing issue, with 57 percent of organizations acknowledging their data is not fit for AI use. Without investment in data governance and preparation, businesses will struggle to meet their objectives and risk reputational damage. Marketers in particular need to ensure the data feeding their campaigns is trustworthy and usable, or risk eroding consumer confidence.
AI agents, meanwhile, have become feasible thanks to advances in multimodal and composite AI. These agents are capable of performing complex tasks with limited human intervention. They offer enormous promise for customer engagement and automation, but organizations remain cautious about entrusting decisions to systems that could make errors or be exploited through security vulnerabilities. Communicators will need to explain both the potential and the risks of AI agents to stakeholders, while shaping policies that protect brand integrity.
This year also marks the debut of AI-native software engineering on the Hype Cycle. Although still emerging, the practice shows how AI is beginning to reshape the very discipline of software development. At present, much of this involves coding assistants and testing tools. In the future, it will mean rethinking engineering roles so that humans can focus on tasks requiring creativity, critical thinking and empathy. For communicators, this signals a broader narrative shift: AI is no longer just a tool layered onto business processes, it is becoming woven into the infrastructure of how organizations function.
The 2025 Hype Cycle underscores that AI is entering a more complex and pragmatic phase. For marketers and communicators, the path forward is not about hype but about building strong foundations, addressing governance and regulatory concerns, and demonstrating how AI contributes to long-term productivity. Those who can frame AI as a disciplined, sustainable capability rather than a passing trend will be best positioned to guide their organizations through this transition.
