AI Agency: Data, Governance, and Costs Hinder Scaling

Leaders describe a bumpy march toward the industrialization of AI agents: only 16% believe their processes are ready, and 15% report multi-agent deployments at scale. By 2030, 74% expect that a significant portion of processes will need to be redesigned, but data foundations (72%), governance (70%), and integration (67%) remain major obstacles. Training and infrastructure expenditures, along with workforce transformation, are already straining budgets.

Organizations Unprepared and Budgets Under Pressure

Only 16% of leaders report that their current processes are ready for agent-based adoption. Employee training costs and token expenses are already putting pressure on budgets, and unforeseen expenditures are anticipated. Leaders are urged to better forecast spending related to infrastructure, employee readiness, AI mastery, and process adaptation for agents. Investing in AI-appropriate infrastructure, a solid database, and training for AI understanding is deemed essential for large-scale deployment. Allocating sufficient resources for workforce transformation is among the priorities. Meanwhile, the pace of technological change exceeds the adaptability of many organizations, which is expected to continue posing challenges. The transformation of processes and workflows is recognized as key, while half of the leaders indicate that their organization is not investing enough in these workforce transformation efforts.

What Needs to Change to Become "Agentic"

Transitioning to an agentic model involves redesigning existing processes and reskilling the workforce to work differently, smarter, and more efficiently. Companies will also need to redeploy employees to less repetitive tasks, restructure their organizational and financial models, recover overlooked value opportunities, and recalibrate their performance indicators around the use of agents. Methodologically, this involves developing an integrated agentic roadmap, treating the layering of agents as a bridge rather than a destination, and defining an operational model that combines humans and agents. This evolution is described as a relational transformation rather than purely technological, and it requires equipping workforce transformation with the necessary resources.

By 2030: Redesigned Processes and Largely Autonomous Agents

By 2030, 74% of leaders believe that about half of processes will be rethought or recreated around AI agents. 61% expect that the majority of their processes will be handled by agents, which will operate largely autonomously, with little or no human intervention. Leaders anticipate new roles and use cases where humans and AI co-create value. Nearly half, 43%, foresee a major disruption of jobs as routine and structured tasks become autonomous and are taken over by agents.

Current State: From Experimentation to Limited Orchestration

According to a survey of 501 leaders involved in AI, the sector is transitioning from experimentation to production deployments. Most organizations still struggle to build an integrated and executable roadmap. 42% are testing a small number of agents, 43% are expanding deployments across functions, but only 15% report orchestrated multi-agent deployments at scale, primarily in customer service, IT, and engineering.

Leaders' Vision: Rethought Models and Governance to Build

Leaders are questioning the viability of their business models and the impact of AI agents on processes and outcomes. Nearly two out of three leaders are reconsidering their business model, and half claim to already have a clear vision of their future operational model based on agents. To scale up, 72% cite the lack of a unified and accessible database, 70% mention the difficulty in trusting and governing agents, and 67% point to the cost and complexity of integration.