Enthusiasm breeds execution and results. This is usually an equation that describes the integration of technology into an enterprise. Agent AI is no exception. The trends observed in The Prompt Economy over the past week show that companies are passionate about execution and building systems to drive results.
Enthusiasm for agent AI far exceeds organizations’ readiness, according to a new report from Harvard Business Review Analytic Services. Most executives expect agent AI to transform their businesses, and many believe agent AI will become the norm across industries. Early adopters are already seeing improvements in productivity and decision-making. However, real-world use remains limited for most organizations. According to the report, only a minority are using agent AI at scale, and many struggle to translate high expectations into consistent business outcomes.
This gap is not a question of trust in technology, but of readiness. The report shows that while data infrastructure is improving, governance, workforce skills and clear measures of success lag. Few organizations have defined what success looks like and how to manage risk when AI systems operate more autonomously. Leaders making progress tend to focus on practical use cases, invest in workforce readiness, and tie agent AI efforts directly to business strategy. The report concludes that agent AI can provide meaningful value, but only to organizations willing to rethink processes, invest in talent, and put strong guardrails in place before scaling.
“The gap between expectations and reality remains wide,” the report said. “If your organization is prepared, your implementation is more likely to be successful and you can close the gap.”
Singapore standards
Governance can also be mandated. According to Computer Weekly, Singapore has introduced what it claims is the world’s first formal governance framework designed specifically for agent AI. The framework, announced by the country’s Minister of Digital Development and Information at the World Economic Forum in Davos, aims to help organizations deploy AI agents that can plan, decide and act with limited human input. Developed by the Infocomm Media Development Authority (IMDA), the framework builds on Singapore’s earlier AI governance efforts, but shifts the focus from generative AI to systems that can perform real-world actions such as updating databases and processing payments. The goal is to balance increased productivity with protection against new operational and security risks.
The framework provides practical steps for enterprises to set clear limits on AI agent autonomy, define when human approval is required, and monitor systems throughout their lifecycle. It also highlights risks such as fraud and automation bias when people place too much trust in systems that have worked well in the past. Industry leaders welcomed the move, saying clear rules are needed as agent-based AI begins to influence decisions that impact the real world. IMDA positions this framework as a living document and is seeking feedback from companies as it continues to refine its guidance for testing and monitoring.
Advertisement: SCROLL TO CONTINUE
identity elements
Another report warns that while companies are rushing to adopt agent AI, they are lagging behind when it comes to governance and security. Accenture and Okta executives say most companies already use AI agents in daily operations, but few have implemented effective monitoring. According to Okta, more than 9 out of 10 organizations are using AI agents, but only a minority believe they have a strong governance strategy. Accenture’s research points to a similar imbalance, showing that AI agents are being widely used without a clear plan to manage the risks they introduce.
The report argues that a central challenge is that AI agents are increasingly behaving like digital employees, rather than being managed as such. These agents require access to systems, data, and workflows to function, creating new risks if their identities and permissions are not clearly defined. The authors recommend treating AI agents as formal digital identities, with clear rules for authentication, access, monitoring, and lifecycle management. Without this structure, organizations risk unmanaged “identity sprawl,” where agent AI goes from being a productivity boost to a major security and compliance issue.
“Agents need a unique identity,” the report states. “Once you accept that, everything else flows, including access control, governance, auditing, and compliance.”
