新兴的智能代理企业：领导者如何驾驭人工智能新时代

by Sam Ransbotham, David Kiron,Shervin Khodabandeh, Sesh Iyer,and Amartya Das AUTHORS Sesh Iyer is a managing director and seniorpartner at BCG and the North America chairfor BCG X, where he helps clients drivelarge-scale AI transformations. He is the Sam Ransbotham is a professor of analytics atthe Carroll School of Management at BostonCollege, as well as guest editor forMIT SloanManagement Review’s Artificial Intelligence David Kiron is the editorial director, research,ofMIT Sloan Management Reviewand programlead for its Big Ideas research initiatives. Amartya Das is a principal at BCG andcurrently serves as an ambassador at theBCG Henderson Institute, where he leadsresearch on the impact of technology andAI on society, focusing on how emerging Shervin Khodabandeh is a managing directorand senior partner at Boston Consulting Group(BCG) and the coleader of its AI businessin North America. He is a leader in BCG X CONTRIBUTORS Todd Fitz, Kevin Foley, Sarah Johnson, Matt Langione, Michele Lee DeFilippo, Amanda Luther,Jennifer Martin, Samantha Oldroyd, Meenal Pore, Allison Ryder, Taylor Zhang, Leila Zhu,Leonid Zhukov, David Zuluaga Martínez The research and analysis for this report was conducted under the direction of the authors as partof anMIT Sloan Management Reviewresearch initiative in collaboration with and sponsored by S. Ransbotham, D. Kiron, S. Khodabandeh, S. Iyer, and A. Das, “The Emerging Agentic Enterprise:How Leaders Must Navigate a New Age of AI,”MIT Sloan Management Reviewand Boston SUPPORTING SPONSORS 1Introduction4Strategic Tensions When Adopting Agentic AI11A Strategic Overhaul of Workflows, Governance,Roles, and Investment17Conclusion18Appendix Introduction Executives have long relied on simple categories to frame how technology fits intoorganizations: Tools automate tasks, people make decisions, and strategy determineshow the two work together. That framing is no longer sufficient. A new class of sys-tems — agentic AI — complicates these boundaries. These systems can plan, act, andlearn on their own. They are not just tools to be operated or assistants waiting for For strategists, agentic AI’s dual nature as both a tool and coworker creates newdilemmas. A single agent might take over a routine step, support a human expertwith analysis, and collaborate across workflows in ways that shift decision-makingauthority. This tool-coworker duality breaks down traditional managementlogic, which assumes that technology either substitutes or complements, auto- The separation of technology and strategy inside most organizations exacerbates thischallenge. Technology executives focus on technology issues, making pilot, vendor,or infrastructure decisions. Strategic executives focus on markets, competition, and ABOUT THE RESEARCH This report presents findings from the ninth annual global research study onartificial intelligence and business strategy byMIT Sloan Management Review and Boston Consulting Group. In spring 2025, we fielded a global survey andsubsequently analyzed records from 2,102 respondents representing more than21 industries and 116 countries. We also interviewed 11 executives leading AI Our research examines the speed of agentic AI adoption inside organizations, whichhas outpaced the adoption of traditional and generative AI. Exploring how agenticAI relates to capital and labor, this report outlines a series of resultant tensions andoffers suggestions for organizations seeking to resolve them. A Tidal Wave of Adoption,a Trickle of Strategy models. Our research, based on 2,000-plus respondentsto a global survey and interviews with leading executives,finds that organizations have multiple options for obtain-ing value from agentic AI. It offers possibilities not only Despite the technology’s wide-ranging implications,organizations are rapidly adopting agentic AI, well beforethey have a strategy in place. Their adoption of traditionalAI has climbed to 72% over the past eight years, accord-ing to our survey. (see figure 1.) Generative AI achieved70% adoption in just three years. In just two years, agenticAI has already reached 35% adoption, with another 44% The Executive Dilemma Agentic AI’s dual nature as both a tool and coworkercreates competing organizational pressures that man-agement frameworks cannot resolve. IT leaders want The result is a growing strategic risk: Agentic AI is spread-ing across enterprises faster than leaders can redesignprocesses, assign decision rights, or rethink workforce FIGURE 1Traditional AI Adoption Since 2023, the percentage of organizations piloting ordeploying AI solutions has risen 22 percentage points. Organizations and individuals can use agentic AI to dif-ferentiate themselves because they can now take advan-tage of the ways that agentic AI does not fit traditionalmanagement frameworks. The same system that offerscost reductions through tool-like automation also enables performance management frameworks and supervisionprotocols