TLDR: A new study by the IBM Institute for Business Value, ‘AI Projects to Profits,’ indicates that businesses globally are rapidly integrating AI agents into core operations, expecting an eight-fold increase in AI-enabled workflows by the end of 2025. Executives view AI agents as crucial for enhancing decision-making, improving efficiency, and driving significant business value, moving beyond mere experimentation.
A groundbreaking study from the IBM Institute for Business Value, titled ‘AI Projects to Profits,’ highlights a pivotal shift in how enterprises perceive and implement artificial intelligence. The research, which surveyed 2,900 executives worldwide, reveals that AI agents are no longer considered experimental tools but are now seen as indispensable for future business operations.
The study projects a dramatic increase in AI-enabled workflows, anticipating an eight-fold surge from 3% today to 25% by the close of 2025. This rapid adoption underscores a growing reliance on agent-based AI for improved decision-making and automation across various sectors.
Key findings from the report include:
Budget Allocation: A significant 64% of AI budgets are currently being allocated to central business units, indicating a strategic integration of AI into core functions.
Decision-Making: Over two-thirds (69%) of surveyed executives identify ‘improved decision-making’ as the most crucial benefit derived from agentic AI systems.
Efficiency and Performance: By 2026, an overwhelming 83% of respondents expect AI agents to enhance process efficiency and overall performance.
Adaptability: Furthermore, 71% of executives believe that AI agents will possess the capability to autonomously adapt to evolving workflows.
Organizational Future: A substantial 70% of surveyed executives consider agent-based AI vital for the future trajectory of their organizations.
Francesco Brenna, VP & Senior Partner, AI Integration Services, IBM Consulting, emphasized this transformative potential: “We are seeing more and more clients see agent-based AI as the key to helping them go beyond incremental productivity gains and actually realize business value from AI, especially when applied to their core processes such as supply chain and human resources.” He further clarified that this evolution is not about simply plugging an agent into an existing process, but rather ‘re-architecting how the process is executed, redesigning the user experience, orchestrating agents end-to-end, and integrating the right data to provide context, memory, and intelligence throughout.’
The IBM ‘Guide to AI Agents’ further elaborates on the foundational components of these intelligent systems, which include perception (structuring inputs), planning (breaking down goals), memory (short and long-term), reasoning/decision-making, action/tool-calling, communication (human and agent-to-agent), and learning & adaptation. These elements collectively form an ‘observe–orient–decide–act’ feedback loop, drawing parallels from robotics.
Architecturally, IBM spotlights three core patterns: Vertical/Hierarchical for sequential workflows, Horizontal/Peer-to-Peer for collaborative tasks like brainstorming, and Hybrid/Dynamic Leaders as a practical middle ground where leadership adapts to project needs.
However, the report also stresses the critical importance of robust governance. Autonomous decision-making inherently amplifies AI risks such as bias, drift, and ‘jailbreaks.’ IBM’s recommendations for mitigating these risks include implementing ‘guardrails-as-code’ within orchestration layers, conducting sandbox and simulation testing before production deployment, ensuring real-time observability for accuracy and context, and integrating ‘hard kill-switches’ for runaway agents.
Also Read:
- Enterprise AI Agent Orchestration: Rapid Growth and Maturation in Late 2025
- Google Cloud Report: Over Half of Enterprises Now Deploying AI Agents, Driving Accelerated ROI
While media narratives often hail 2025 as ‘the year of the AI agent,’ the IBM study provides a grounded perspective, confirming widespread exploration and development of AI agents by developers, with 99% of those building enterprise AI applications actively engaged in this area. This indicates that while hype exists, the underlying industry movement towards agentic AI is substantial and strategic.
