技术导航：第二版 代理型AI在行业中的应用

Knowledge Institute CONTENTS Executive summary4Unified operational control in oil and gas6Inventory discrepancy in the retail industry12How to manage discretionary risk in financial services20Agentic AI is redefining the manufacturing value chain30How agentic AI is re-engineering telecom business sales44 In our last edition of Tech Navigator, weexplained the rise of agentic AI and howit amplifies business. AI agents are alreadyenabling organizations to reimagine andre-engineer business processes — whenthe ROI is justified — and on a smaller scale,drive process automation and simplification.Beyond re-engineering, ongoing experimentspave the way for vibe coding and explorationfor business users to continuously innovate. On the technology front, building AI runwaysthat power platform and democratizationcapabilities across all layers of the architectureis proving key to effective AI scaling. Agentic AI is already proving its value in broaddomains, such as IT, business operations, andinformation intensive business processes.However, in this new Tech Navigator report,we shine a spotlight on how Infosys is usingagentic AI to solve real-world challenges infive distinct industries. Many large enterprises have found thatimplementing intelligence at scale is easiersaid than done. They need an architecture-first approach, and they need to treat AI as atransformative program. At the same time,it is important to set up the right hub andspoke operating model, backed by a value-and outcome-driven AI investment approach. Agentic AI now has the power to transformmission-critical domains: unified operationalcontrol in upstream oil and gas companies;inventory discrepancy in consumer, retail,and logistics organizations; discretionary riskmanagement at global banks; aircraft and shop-floor maintenance in manufacturing;and business sales process in telecomcompanies. In so doing, we take the readeron a whistle-stop tour of what’s really neededto take agentic AI from a point solution to thenexus of digital and AI transformation. But technology isn’t everything.Highly autonomous, most agentic AIimplementations also rely on humansfor governance and control. Harnessingthe requisite talent is therefore not justimportant but critical to any agentic AIchange management strategy. Organizationsshouldn’t just hire more data scientists and AIexperts but look at how this new technologywill redefine current roles and require trainingin the continuous, adaptive oversight ofagentic AI systems. The goal? Higher levels of efficiency,adaptability, creativity, and business insight,all needed in a world where threats multiply— from economic volatility to supply chaindisruptions. Both pragmatic and highly relevant, thisedition has been designed to give insightinto how agentic AI could rise to be one ofthe most important tools in an organization’sstrategic arsenal — depended on byexecutives and consumers alike. Unified operationalcontrol in oil and gas Agentic AI does not always play well with traditional, sometimes antiquated, software that is often critical tooperations in asset-heavy industries. However, the new model context protocol forms a bridge between thesesystems, maintaining unified operational control while allowing for the synthesis of real-time information. Oil and gas industry executives increasinglyagree that artificial intelligence is essentialfor understanding the vast data generatedby Internet of Things (IoT) devices thatare multiplying in asset-heavy industries.Advanced technologies can harness thisdata to optimize operations and enhancedecision-making. In fact, 59% of global oiland gas executives recently told Infosys thatAI will make a significant contribution totheir revenue in three years. And 75% saythat AI investments will deliver a measurablecompetitive advantage in this same timeframe. At Infosys, we’re supporting our clients’use of AI across the energy value chain —embedding the technology into oil fields,drilling rigs, compressor stations, anddistribution networks, among other criticalupstream, midstream, and downstreamresources (Figure 1). As a result, these smartassets lead to gains in efficiency, resiliency,and operational integrity. And through theuse of generative AI, IT and operationaltechnology (OT) data streams provideinsights into plant configurations in real time,reduce the downtime of vessels, and supportmaintenance planning. Further downstream, model context protocol (MCP) to query realtime and historical data from proprietarysoftware management systems, leading toautonomous, efficient, and fail-safe upstreamoperations. AI creates a smarter workplace throughpersona-based copilots, including personasin smart inspections and field operations,marketing, and energy trading. In this chapter we concentrate onupstream operations — as shown on theleft side of Figure 1 — which includesexploration, drilling, and extractionof crude oil and natural gas. We lookspecifically at how agentic AI can use the A