New York – September 24, 2026 -- Direct AI-model usage accounts for just 9.4% of an AI agent's recurring monthly operating cost, according to an illustrative model cited in a new report from The Conference Board, underscoring how governance, monitoring, and human supervision drive the bulk of expenses in agentic AI deployments.
Executives rank AI as the top investment priority for 2026
In The Conference Board's C-Suite Outlook 2026, 43.6% of executives identified AI and technology as their leading investment priority, ahead of any other category. The new report argues that capturing value from that spending requires more than purchasing tools or running pilots.
A five-stage framework redesigns work around AI agents
The framework, based on interviews, focus groups, and research with senior HR, talent, and AI leaders, instructs organizations to first define the business outcome they want to improve before deploying agents. Companies are then told to break workflows into individual tasks, assigning each to AI, humans, or a combination based on quality requirements, accountability, and error consequences.
"AI agents can complete more and more work, but capability alone doesn't create value," said Allan Schweyer, Principal Researcher, Human Capital, The Conference Board. He said companies need to know what outcome they're targeting, what agents should and shouldn't do, and whether the redesigned process actually performs better.
Report warns against counting freed-up time as automatic financial gain
The framework's fourth stage calls for measuring what people and agents produce together rather than simply confirming a task was completed. It recommends distinguishing three categories of gains: cash savings from disappearing costs such as overtime or contractor fees; avoided future costs from planned spending or hiring no longer required; and higher-value work, where redirected employee time produces a separately measurable result.
Leaders must decide how verified gains are redistributed
In the fifth stage, the report says verified gains can be used to cut costs, absorb more work without proportional hiring, fund reinvestment, or benefit employees and customers. It states that leaders should communicate clearly how those gains will affect jobs and staffing.
Report flags risk to early-career skill-building pipelines
The report warns that automating routine tasks can eliminate the work through which employees traditionally build judgment and experience. It recommends organizations deliberately replace that learning through mentoring, job rotations, supervised practice, and progressively complex assignments.
Conference Board says HR and IT must share ownership of agentic AI
"Agentic AI is not simply an IT initiative, nor is it solely an HR initiative," said Diana Scott, US Human Capital Center Leader, The Conference Board. She said business leaders must own the outcome, technology must provide secure tools, finance must test the economic case, and HR must ensure roles, skills, and staffing evolve alongside the technology. The research was produced in collaboration with SkillsRight, a nonprofit focused on workforce strategy and skills-first talent systems.