代理操作的蓝图如何打造互联企业（英）2026

How to build an interconnected enterprise Contents Paradigm shift...................................................4 Unlock potential..............................................18 Action guideRebuild for autonomous execution................20 Appendices......................................................27 Key Agentic AI operating modelsare already here. 55% of businesses areactively developing ordeploying an agentic AI Businesses developing or using these modelsexpect agentic AI to dramatically improve outcomesby the end of 2027: +45% operational efficiency, Disconnected workflows obstruct Fragmentation—not talent or AI maturity—is thebiggest constraint: 82% of leaders say siloedfunctions block the value autonomy delivers. 77% of leaders are investing in data quality andgovernance improvements to enable autonomous Digital twins are tomorrow’s 72% of executives plan to implement real-time simulation and scenario analysis at the Autonomous workflowadoption accelerates 5.4 times Their effect is multiplicative, not additive. Changemanagement readiness and AI governance are Introduction Your operating modelhas become a liability Enterprises are discovering an uncomfortabletruth. Operating models they spent decadesoptimizing have become the primary barrierto AI value. The problem isn’t talent. It isn’t AImaturity. It’s structure. Standalone functions— Future enterprises operate drastically differently.They organize around interconnected workflowsthat allow AI to continuously execute, optimize, AI agents manage operational execution whileexperts focus on critical thinking, judgment,and strategic oversight. AI is no longer a bolt According to new research from the IBMInstitute for Business Value (IBV), 82% ofC-suite executives say functional silos block This new operating paradigm is already in flight.55% of organizations are actively developing ordeploying an agentic AI operating model. 60%plan to adopt next-generation delivery structureswhere AI agents coordinate integrated workflows This report goes far beyond diagnosing status-quo problems, however. It details a new AI-firstoperating model organized around agenticworkflows spanning functional domains. Itsdata-based insights are intended for thoseresponsible for enterprise business operations:the CFO, COO, CHRO, CIO, CSCO, CCO, and As AI agents grow capable of executinghorizontal workflows, processes designed arounddepartments become costly. Data locked insidefunction-specific architectures cannot powerenterprise-wide intelligence and AI autonomy. Based on an IBV survey of more than 2,000 C-suite executivesacross 16 countries and 17 industries, this report identifiesboth the biggest opportunities and main obstacles to becoming Part One outlines why AI has outgrown traditional deliverystructures and introduces the agentic model reshaping enterpriseexecution. Part Two details the key pillars required for autonomousoperations: interconnected workflows, interoperable data, andenterprise-wide orchestration. Part Three showcases how a cohort Welcome to autonomous business operations. Welcome to the 75%of executivesagree that AI willsignificantly redefine The interconnected enterprise, defined An operating model organized around agentic workflows spanning functionaldomains. Data interoperability across the enterprise lays the foundation forboth specific workflows and an orchestration layer enabled by a digital twinthat mirrors operations. In the interconnected enterprise, AI agents execute Paradigm shift Today most enterprises are designed around functionalspecialization—finance, procurement, supply chain, HR,technology—in tune with the demands of analog work. Coordination has historically required manual approvals, spreadsheets,and large teams. Functional departments used to be the most practical wayto manage complexity. But there are downsides to that approach: handoffs, Rapid maturation of AI makes these drawbacks more glaring. Fragmentedworkflows force AI systems to automate isolated tasks rather than driveoutcomes. This leaves structural gaps between functions—and the work AI agents will begin crossing functional boundaries even as old systemspersist. Leaders are faced with a choice: control change or react to it. In ourresearch, 59% believe agentic AI will bridge organizational silos by the end An AI-first enterprise requires workflows engineered for continuousorchestration and teams aligned to outcomes, not departments. In this context,the old categories of shared services, outsourcing, and internal operations arenot relevant. 77% of survey respondents say transforming GBS is a strategic At the core of the new operating model are interconnected workflows thatcut across functions, locations, and partners. 74% of executives say AI isdissolving traditional geographic and functional boundaries. Our researchshows that leaders expect cross-domain workflow teams to outperform An AI-first enterprise requires workflowsengineer