《Coresight Research&英特爾：2025年零售技術十大趨勢報告（英文版）（21頁）.pdf》由會員分享，可在線閱讀，更多相關《Coresight Research&英特爾：2025年零售技術十大趨勢報告（英文版）（21頁）.pdf（21頁珍藏版）》請在三個皮匠報告上搜索。

1、January 6,2025Copyright 2025 Coresight Research.All rights reserved.Retail 2025:10 Trends in Retail TechnologySponsored byExecutive SummaryIn-StoreE-CommerceOperations1.Retail Computing:On the Edge and in the CloudThe center of computing has become decentralized,taking place both in the cloud and on

2、-premisesi.e.,on“the edge”of the networkdriven by an explosion of data,the proliferation of in-store devices and the requirements of computer vision(CV).2.Computer Vision:The In-Store Swiss Army KnifeCV is fundamental to easing friction for consumers and enabling many loss-prevention and inventory-t

3、racking applications.The number of CV applications is likely to increase further in 2025 due to the ubiquity of in-store cameras and the growing number of capabilities and applications powered by AI(artificial intelligence)and CV.3.Mobile Devices Empower Frontline WorkersRetailers face challenges fr

4、om a high turnover rate and growing competition for associates from other sectors.Mobile devices provide one possible solution as these devices help train associates,make them more efficient and assist them in providing better customer service,creating a better work environment.4.The Store:Retail Me

5、dias Final FrontierExpanding retail media into the store presents a large and lucrative opportunity that retailers should consider now in order to stay ahead of the competition.5.Shopping Nirvana:Images,Voice and Natural LanguageCurrently,various technologies for image,voice and natural-language sea

6、rch exist.In 2025,these technologies will be increasingly combined,with AI integration leading to superior customer experiences and more sophisticated and seamless offers.6.Personalization at Scale Could Finally Be HereToday,customer data platforms(CDPs)can identify unknown customers and enhance seg

7、mentation and targeting.However,moving forward,GenAI(generative AI)will be able to create personalized,relevant messages and communications based on the customer profiles provided by CDPs.7.Democratize Data Access&AnalysisOften,enterprise data are siloed and unstructured.GenAI enables everyone in an

8、 enterprise to access data that are stranded in siloes and in unstructured formats via natural language.8.Agentic AI Is HereSeveral major enterprise software providers have announced the availability of AI agents,as well as the tools required for creating them,meaning that AI agents are likely to se

9、e increased deployment in 2025 and beyond.9.GenAI Can Unlock Supply ChainsGenAI can acquire unstructured data from the entire supply chain and allow users to ask questions using natural human language.As we move forward,we expect AI agents will increasingly manage unplanned supply chain events.10.De

10、mand Forecasting 3.0Advanced demand forecasting tools leverage ML(machine learning)and external data to create highly accurate forecasts.Now,GenAI offers the ability to incorporate unstructured data to improve forecasts and to query forecasts using natural language.Coresight Research AnalysisWe pres

11、ent 10 key technologies that we expect to be influential in retail in 2025,including discussions of what we think is significant today and what will be important in the year ahead.What We ThinkRetailers and brands are currently in the enviable position of having several powerful technologies to choo

12、se from to make their businesses more productive and efficient,as well as to entrench themselves with their customers.AI appears practically everywhere throughout our list of retail technology trends,from computer vision and visual shopping and search to data analysis and personalization.Building on

13、 this potential,GenAI promises to bring the power of AI to everyone,and,though in the early stages,AI agents will be able to act on our behalf,bringing even more productivity gains and new capabilities to the masses.Introduction.4Retail Technology:Market Outlook.5Directional Outlook.5Sector Outlook.

14、5Tailwinds and Headwinds.6Pervasiveness of AI.610 Trends in Retail Technology for 2025:Coresight Research Analysis.7In-Store.8E-Commerce.12Operations.14What We Think.18Implications for Brands/Retailers.18Implications for Technology Vendors.19Impacts from AI.19Notes.20About Sponsored Reports.21About

15、Intel.21ContentsIntroductionUndoubtedly,2025 promises to be an exciting year for retail technology,as retailers and brands have a growing selection of technologies to choose from,with many technologies benefiting from advances in AI(artificial intelligence).At this point,many retailers have moved be

16、yond the experimentation stage and have deployed GenAI(generative AI),further improving via accelerators,productivity enhancers and new capabilities.For those unwilling to get stuck in with large language models(LLMs),there are a myriad of applications available that enterprises can use to do the he

17、avy lifting.Moving forward,AI agents promise to bring a new wave of AI excitement,as these autonomous,intelligent programs can act on our behalf and even communicate and interact with other agents.This report,part of our Retail 2025 series,delves into the 10 key technologies that we expect to be inf

18、luential in retail in 2025,including AI,GenAI,computer vision(CV)and more.It is made available to non-subscribers of Coresight Research through its sponsorship by Intel.4IntroductionThe climate for retail tech spending remains largely positive,driven by the need for retailers to invest in tools to r

19、emain competitive in a relentlessly competitive sector.Positive retail sales growth acts as an enabler for investment,and the current,high level of retail-tech innovation is producing numerous new-and-improved applications and platforms.Retail Technology:Market OutlookDirectional OutlookWe expect th

20、e following technology categories to receive a relatively greater share of retail-tech investments:AI and GenAIWe are clearly in a period of great excitement around AI,driven by the advent of GenAI and continued demonstrations of AIs powerful benefits.Now,more than two years after the watershed laun

21、ch of ChatGPT 3.5,enterprises are benefiting from a flourishing landscape of AI models,platforms,applications and agents,in addition to AI-driven tools that can help create AI applications.Moreover,retailers and enterprises can take advantage of multiple layers of AI,including automation,algorithms,

22、machine learning(ML),GenAI and AI agents to solve their business problems and achieve their business goals.AI is also a foundational technology for CVwhich continues to gain new features and flexibilitya technology that can perform many in-store functions for consumers and retailers alike,greatly en

23、riching the shopping experience.Edge ComputingAI is bringing some computing back to retail premises,though it is different from the premises-based computing of old.Edge devices and servers now contain processors with AI cores and GPUs,allowing data and video processing to be performed within stores

24、at quicker and more cost-effective rates than those seen when sending data to the cloud.Supply Chain TechnologiesDespite the Covid-19 pandemic sitting firmly in the rear-view mirror,supply chains have not returned to their pre-pandemic configurations and remain as challenged as ever due to commodity

25、 prices,shipping disruptions and the slow recovery(or volatility)of the supply of key inputs.Rapid changes in customer preferences and shopping habits add further variability to supply chains.New technologies,such as tags for items and containers that can provide location data and GenAI,can fill in

26、the gaps.For example,GenAI can make sense of unstructured supply chain data,while AI/ML is a core technology behind demand forecasting and inventory and pricing optimization.Sector Outlook5Retail Technology:Market OutlookAI,ML and GenAI underlie our top 10 tech trends due to their ability to analyze

27、 large amounts of data to find relationships,anomalies and opportunities in an automated and robust fashionmany of these opportunities may not be visible to the human eye or would require years of analysis to find.GenAI,in particular,brings an interface that utilizes natural human language to analyz

28、e the data,enabling anyone to perform ad hoc analysis:data analysis is no longer only relegated to data scientists and does not require hours of programming or creating dashboards.In addition to data analysis,GenAI enables the creation of text,images,video,audio,synthetic data,and,perhaps most impor

29、tantly,computer code.Coresight Research has developed our CORE Framework for AI in retail to give retailers and brands a head start in strategizing around AI deployment.The framework classifies the applications of AI within four main categories that can help retailers win new business,run their oper

30、ations more efficiently and create strong relationships with their customers.Sustainability occupies the center of the framework,as efficiency and productivity gains mean the consumption of fewer resources.We view AI as a fundamental technology that can improve all of a retailers activities,and we e

31、nvision its application across all the trends outlined in this report.Pervasiveness of AI Continued economic and retail sales growth AI and app fatigue,which are driving enterprises toward modern data structures and comprehensive platforms Successful demonstrations of AI use cases and ROI(return on

32、investment)High-margin revenue from new sources,such as retail media,selling data and licensing technology Tailwinds Pressure on consumer budgets from the absolute level of prices,though inflation has subsided Retailers reluctance to make large bets on technology Spending on AI,which could starve ot

33、her areasHeadwindsSource:Coresight ResearchFigure 1.The Coresight Research CORE Framework for AI in RetailTailwinds and Headwinds6Retail Technology:Market Outlook10 Trends in Retail Technology for 2025:Coresight Research AnalysisIn Figure 2,we summarize our top 10 retail technology trends for 2025,c

34、ategorized into three areas:in-store;e-commerce;and operations.We delve into each trend below.Figure 2.10 Trends in Retail Technology for 2025Source:Coresight ResearchOperations7.Enterprise data8.AI agents9.GenAI-powered supply chains10.Demand forecastingE-Commerce5.Shopping nirvana6.Personalization

35、In-Store1.Edge computing2.Computer vision3.Mobile devices4.Retail media710 Trends in Retail Technology for 2025:Coresight Research AnalysisWhat to Know NowIn-StoreFigure 3.In-Store Devices That Can Compute on the Edge or in the CloudSource:Coresight ResearchRetail stores are replete with many device

36、sPOS(point-of-sale)terminals,associate mobile devices,shelf-edge cameras,robots,displays,overhead cameras and traffic counters,among othersand this list continues to grow as new applications are invented.Many of these devices,such as cameras that operate in near real time,require a speedier response

37、 than that provided by sending data to the cloud,processing it there and awaiting a reply(i.e.,they require lower latency).There are also opportunities to deploy AI within retail stores,using the technology for personalization,managing retail media networks(RMNs),clienteling,interactive experiences

38、and CV applications.All of these applications require in-store computing,in the form factor of an edge device and an on-premise edge server.CV,in particular,requires a substantial amount of resources in terms of latency requirements and data generation.Video files are often large(transmitting 1080p

39、video requires a data rate of five megabits per second(Mbps)and require significant computing resources,whereas on-premises systems represent more of a fixed cost;for instance,the Dell PowerEdge XR7620 server,which is suitable for AI and ML,retails for about$12,000 and runs on the 4th Gen Intel Xeon

40、 Scalable Processor.To support the needs of AI,current generations of PCs(also called edge devices)and servers now feature AI capabilities.For example,edge devices with new Intel Core Ultra processors integrate dedicated NPUs(neural processing unitsi.e.,AI functions)and enhanced built-in GPUs,which

41、can share the processing load with the CPU and are designed for power efficiency,performance,integrated graphics and AI capabilities.The server displayed above contains 12 processor cores in addition to supporting AI-related features,such as matrix calculations and accelerators for deep learning.Ven

42、dors of edge computing hardware include various leading PC providers,such as Dell,HP and HP Enterprise,and Lenovo,as well as specialists such as Supermicro;the products provided by these companies often include chips and processors from the likes of AMD,Intel and Nvidia.Read More on This Trend:Read

43、our Retail Challenges Drive Edge Technology Investment report for an outlook on US retailer IT spending and the categories that retailers expect to have the greatest impact on revenue.Retail Computing:On the Edge and in the Cloud1DiscussionThe center of computing has become decentralized,taking plac

44、e in the cloud and on-premisesi.e.,on“the edge”of the networkdriven by an explosion of data,the proliferation of in-store devices and the requirements of CV.Whats Important in 2025The continuing development of AI and CV will increase computing requirements,and edge computers,or AI PCs and devices,wi

45、ll continue to become more powerful as more AI chips and functions are added.Whereas,previously,the center of computing had largely migrated from on-premises to the cloud,today,we are witnessing a shift to hybrid cloud configurations in order to serve the proliferation of in-store devices and data-r

46、ich applications.This shift allows for a more efficient distribution of computing tasks between on-premises and cloud infrastructurewith each segment doing what it does bestoptimizing performance and responsiveness.In Figure 3,we illustrate the interaction between the cloud and a bevy of edge device

47、s,which may leverage processing on the edge,in the cloud,or both.Dell PowerEdge XR7620 Rack ServerSource:Dell8In-StoreWhat to Know NowCameras and CV play a pivotal role in loss prevention:Cameras located at the entrances and exits of a store can capture photos of shoplifters,while item tags,such as

48、RFID(radio-frequency identification)tags,can identify what was stolen.CV can identify potential shoplifting activities and dishonest transactions such as loitering or shelf sweeps,ensuring the health and safety of customers and associates.Moreover,facial recognition is becoming less controversial as

49、 it sees applications in airline travel,and several retailers are now using the technology to identify known shoplifters and suspects before they can commit theft inside the store.CV offers several opportunities to improve the in-store customer experience.For example,SCOs and smart carts can leverag

50、e CV to scan barcodes and identify produce,in addition to SCOs using CV to determine a purchasers age,decreasing purchase times for orders that contain alcohol or cigarettes.Other CV uses include determining heatmaps for better store layouts and measuring the length of checkout lines to deploy addit

51、ional cashiers during busy times.CV can collect a variety of shopper and in-store data,such as shopper emotions and queue wait times,and measure shopper traffic,enabling the creation of heat maps for optimized store layouts.Read More on This Trend:RetailTech:Loss PreventionUnderstanding the Big Pict

52、ure To Prevent,Mitigate and Recover“Shrink”ing ProfitsComputer VisionThe In-Store Swiss Army Knife2CV continues to find new applications within physical stores due to the ubiquity of cameras and the potential of AI-powered CV.Moreover,new,previously unimagined applications powered by GenAI are just

53、around the cornerthese emerging applications will support various business functions,including inventory tracking,store operations and data collection.DiscussionInventory TrackingLoss Prevention and Associate SafetyEasing Friction for ConsumersOperations and Data CollectionLoss-prevention components

54、 and cameras in a SCOSource:Diebold Nixdorf/Coresight ResearchCameras with and without CV can occupy several locations within the retail store,including in the ceiling,on the shelf edge,in smart shopping carts,mounted on robots that cruise around throughout the store,in handheld devices,inside self-

55、checkout kiosks and even on other locations,such as in automated vending machines and on refrigerator doors.Cameras in the ceiling are a key technology in enabling seamless,or“just walk out,”shopping.Amazon uses a fusion of cameras and sensors in its Amazon Go stores,as opposed to CV-only solutions

56、such as Trigo and the recently defunct Grabango,which offer simpler solutions lacking physical sensors.Recently,Amazon has pivoted toward using its own Dash Cart technology for its grocery stores,despite its prior assertions that its“Just Walk Out”technology was robust enough to handle the associate

57、d complexity.CV is also a core technology in self-checkout kiosks(SCOs),most notably in regard to scanning items;identifying produce;determining purchasers age for products such as alcohol and cigarettes;thwarting customer theft and errors at the kiosk;and reducing transaction times.Shelf-edge camer

58、as,such as VusionGroups Captana and Trax Retails cameras,leverage CV to track the location of items within a store.As shelf-edge cameras are mounted in a fixed position on a shelf,their field of view is limited compared to robot-mounted solutions,such as those from Badger Technologies and Simbe,whos

59、e cameras can traverse the store aisle and swivel to capture a wider field of view.CV is fundamental to collecting data,enabling many loss-prevention and inventory-tracking applications,and easing friction for consumers,thereby improving the shopping experience.Whats Important in 2025The number of C

60、V applications is likely to increase further in 2025 due to the ubiquity of in-store cameras and the growing number of capabilities powered by AI and CV.9In-StoreWhat to Know NowMobile Devices Enable Frontline Workers3Retailers currently face challenges in hiring and keeping associates due to a conf

61、luence of factors.Perhaps most notably,several non-retail sectors,such as restaurants and delivery services,offer higher pay due to recent government legislation,as well as more flexibility in terms of working hours.The retail industry also faces a high staff turnover ratearound 60%annually,accordin

62、g to our analysis of Bureau of Labor Statistics(BLS)datawhich results in higher onboarding,training and recruiting expenses.A high staff turnover rate also results in a steady stream of new associates,making it challenging for customers to see a familiar face when they enter the store.Retail industr

63、y employment stood at slightly below 16 million in October 2024,according to the Federal Reserve Bank of St.Louis,a number that has trended upward since the end of the pandemic and now stands roughly in line with October 2016 levels.However,US retail sales increased by 53%in 20162023,meaning that th

64、e same number of associates is supporting a substantial increase in sales.Doing more with less takes its toll on associates:79%of associates stated that they feel stressed due to the complexity of day-to-day work,according to Zebra Technologies 17th Annual Global Shopper Study,published in November

65、2024.These challenges also take a toll on customer sentiment:70%of consumers stated that it is hard to find an associate to help them while shopping in a store“these days,”per Zebras study.Mobile devices can assist associates with training and operations,enable upselling and improve the employee exp

66、erience,which,ultimately,improves customer sentiment and boosts return visits.Todays mobile devices are much more than simple handheld scanners:they are powerful mobile computers.Employee devices can also boost sales.For instance,associates can add an extra forgotten item to a BOPIS(buy online,pick

67、up in-store)order at pick up,and intelligent devices can include order histories to help associates make relevant suggestions,either on their own or with the help of AI-powered tools that utilize previous purchases,searches,service conversations and other data to offer hyper-relevant recommendations

68、.The latest generation of handheld devices is able to run LLMs,putting the power and the creative abilities of GenAI into associates hands,creating more possibilities for clienteling than ever beforefor example,GenAI could create an image of an outfit for the customer,allowing them to see how variou

69、s items work with each other.The power of these intelligent tools enables new hires to access the customers accumulated history with the retailer,allowing the new associate to act in a similar manner to a long-acquainted personal shopper.In addition to processing transactions,these devices serve as

70、intelligent work hubs for associates,enabling them to consume training videos,look up company policies,make changes to their work schedule and communicate with coworkers via their devices,allowing them to spend more time with customers.Additionally,possessing their own devices makes employees feel m

71、ore valued by their employersthey particularly value being able to set their own schedule:53%of retail managers believe that enabling employees to select their preferred locations,hours and days would contribute to higher employee retention and satisfaction,according to a proprietary Coresight Resea

72、rch survey of mid-to large-sized global employers(conducted in September 2021).Read More on This Trend:Equipping Frontline Workers with More Mobile Devices to Drive Revenue and Delight CustomersDiscussionClienteling app on an associates tablet offering recommendations for customersSource:PredictSpri

73、ngRetailers face challenges from a high staff turnover rate and growing competition for associates from other sectors.Whats Important in 2025Mobile devices provide one possible solution as these devices help train associates,make them more efficient and assist them in providing better customer servi

74、ce,creating a better work environment.10In-StoreAmid the ongoing retail media boomthere are more than 200 RMNs in operation as of November 2024,according to campaign tool provider Mimbithe retail store remains a challenging,albeit lucrative,new frontier.The opportunity remains key as in-store retail

75、 represents about 77%of US retail sales,according to Coresight Research estimates,which means that the in-store revenue opportunity is 4X that of the online opportunity.Screens can be placed in multiple in-store locations,including above deli and bakery counters and on shelves,endcaps,shelving rails

76、,smart carts,cooler doors and SCOs.Electronic shelf labels can also contain ads and promotions.However,putting screens in stores poses significant challenges when compared to online retail media options:First,there is the physical aspectScreens are hardware that must be installed,connected to electr

77、ical power and an ad network,and maintained,repaired and upgraded.This presupposes a certain amount of store infrastructure,including a reasonably fast and stable Internet connection.Second,the online portion of the retail media landscape is already highly complex,comprising retailers,digital market

78、ing platforms,media accelerators,analytics firms and a handful of in-store advertising firms.Adding hardware to this already fragmented sector adds another layer of intermediaries and complexity that must be managed.Challenges of attribution and standardization become even more thorny in the physica

79、l store,as there are few acceptable ways to determine which advertising a customer has viewed.While there are technologies such as traffic counters,cameras and cellphone-tracking technologies,many of these ignite consumer concerns about privacy.These types of concerns have already arisen surrounding

80、 in-store displays on refrigerator doors and the data they could collect on the age,gender and ethnicity of shoppers;additionally,US Congress members have questioned Kroger on its intention to use facial-recognition technology to inform dynamic pricing processes,which they fear could be used to goug

81、e consumers.In-store displays could also include outward-facing cameras,collecting demographic information,as well as presence and dwell times,meaning they could be subject to the same privacy concerns.Smart shopping carts from vendors such as Amazon,GK Software,Instacart(Caper),and Shopic represent

82、 one area of retail media that can be highly personalized and provide attribution.When a customer logs into the cart,it gains access to the customers shopping history and shopping list;AI/ML can then analyze this data to determine the best advertisements to present or those that fit the retailers pr

83、eferences.When the customer checks out,the final purchase list can be analyzed versus the advertisements displayed to determine their effectiveness.Not every retail medium requires in-store screens.For instance,Walmart produces its own in-store radio network,while the company Rockbot offers both aud

84、io and video solutions for retailers.On the other hand,Vestcom produces stick-on price labels that often appear in major US drugstores,with the stores typically receiving a new batch weekly.These tags can feature pricing,product and promotional information,as well as connect to the retailers smartph

85、one app via a QR code,which can provide shoppers with supplemental product information and confirm attribution.Read More on This Trend:Retail-Tech Landscape:Retail Media;all our coverage of retail mediaDiscussionShelf-edge screenSource:VusionGroupShelf-edge promotion and associated app connectionSou

86、rce:VestcomWhat to Know NowRMNs have continued to grow rapidly throughout 2024,offering impressive margins.Whats Important in 2025Expanding retail media into the store presents a large and lucrative opportunity that retailers should consider now in order to stay ahead of competition.The Store:Retail

87、 Medias Final Frontier 411In-StoreMetas Ray-Ban glasses leveraging voice,image recognition and GenAISource:MetaRetailers and consumers are set to benefit from the confluence of three technologies:image recognition and generation,voice recognition and natural-language understanding.Currently,these te

88、chnologies are being combined with the intelligence of large language AI models to create powerful new ways to market products to consumers.Image recognition has been transformed into visual searchthe futuristic concept of snapping a photo from a magazine,catwalk model or person on the street and be

89、ing able to identify and purchase the same item(or at least a similar one)is now here.Basic technologies,such as Google Lens,enable the identification of images and translation of text,allowing users to search for similar images online.Meanwhile,GenAI can create illustrations of objects,complete out

90、fits and appropriate backgrounds on the fly for enhanced brainstorming and shopping.Although voice shopping exists today,it has yet to gain widespread acceptance;however,the technology continues to advance,and voice search could become a must-have application when combined with other leading-edge te

91、chnologies,such as GenAI.Most notably,GenAI makes applications multi-lingual,enabling real-time translation between languages without requiring intermediate translation into English.Voice recognition-powered GenAI chatbots offer the opportunity to communicate in an interactive and flexible manner,wh

92、ich feels more like a genuine human conversation,while simultaneously utilizing GenAIs capabilities to analyze data and create content.Meta has provided a view of the future with its ability to leverage several proprietary technologiesaugmented reality,voice recognition and LLMsvia its Ray-Ban smart

93、 glasses.During a demo at Shoptalk Spring 2024,a speaker from Meta demonstrated a spoken conversation with the glasses as they sought recommendations on an outfit for an event.Later,the Meta Connect 2024 keynote featured a similar demonstration,with the glasses providing an analysis of a live video

94、of a rack of clothing based on the users verbal questions,as depicted in the image below.Metas Orion augmented-reality(AR)glasses add the ability of the wearer to view images generated by the device.Read More on This Trend:RetailTech:Computer VisionApproaching the Holy Grail of Ubiquitous Visual Sea

95、rchDiscussionWhat to Know NowThere are currently several AI-based technologies for image,voice and natural-language search.Whats Important in 2025In 2025,these technologies will increasingly be combined,with AI integration leading to more sophisticated and seamless offers.Shopping Nirvana:Images,Voi

96、ce and Natural Language5E-Commerce12E-CommerceSource:Coresight ResearchPersonalization at Scale Could Finally Be Here6Personalization has long been a dream of marketers,yet this dream has largely remained elusive for a multitude of reasons.Although it would be foolish to declare that long-imagined p

97、ersonalization is just around the corner,there are technologies available today,such as CDPs,that can move marketers in the right direction.AI/ML can be used to personalize product recommendations online based on available data such as customer information,purchase history,web browsing history and s

98、ervice conversations.GenAI can then be used to compile and enhance those recommendations to better resonate with the shopperfor example,creating looks and outfits.(This technology is applicable to the store,too,offering personalization via mobile devices.)CDPs offer a meaningful step toward enabling

99、 personalization.To offer personalization,the CDP must identify the person.It first performs a key functionidentity resolution:piecing together the data“crumbs”that consumers leave behind across a variety of hardware and software platforms when they believe that they are staying anonymous or are in

100、just too much of a hurry to log in or enter their personal information.Once individuals have been identified,this information can be used for segmentation,targeting and marketing analytics.CDPs perform a variety of tasks in the marketing process,as shown in Figure 4.Looking ahead,marketers can combi

101、ne virtually unbounded computing power to analyze data to define ever-smaller customer segments,ultimately arriving at a segment of one person.Yet,composing personalized communications such as emails,websites and text messages is not scalable and would,therefore,be impossible to accomplish for a sin

102、gle employee.GenAI is able to incorporate data such as purchase history,browsing history,website activity and service conversations to create personalized communications in the form of text and media(such as audio,images and video),ultimately creating communications that do not sound computer-genera

103、ted.In addition to messages,these communications can include personalized webpagescomplete with copy and imagesproduct recommendations and product discussions.Thus,the combination of computing power and GenAI could finally accomplish the long-sought goal of personalization at scale.Read More on This

104、 Trend:Customer Data Platforms:Unearthing Buried Treasure in Unified Shopper ProfilesDiscussionWhat to Know NowCustomer data platforms(CDPs)can identify unknown customers and enhance segmentation and targeting.Whats Important in 2025Moving forward,GenAI can create personalized,relevant messages and

105、communications based on the customer profiles provided by CDPs.Figure 4.Key Functional Areas of CDPsIdentification/Acquisition IntegrationAnalysisActivation and Orchestration13E-CommerceDemocratize Data Access&Analysis7In the past,data remained within its respective department:sales data was managed

106、 by sales,financial data remained within the finance department,customer data remained within the marketing department,and so on.An analysis of departmental data required a request to a data-science team or programming knowledge to create a data dashboard,which required a great deal of planning and

107、effort,limiting creativity and serendipity.GenAI democratizes access to data,as its interface is in natural human language,i.e.,humans can ask questions that pertain to an enterprises data with no programming ability required.Thus,data analysis moves out of the hands of data scientists into the hand

108、s of anyone and everyone in an enterprise.Moreover,GenAI makes data analysis faster and more interactive,meaning that executives can quickly receive answers to their follow-up questions.Questions that GenAI can answer include metrics and goals,such as“how do I increase sales?”When asked this,the LLM

109、 parses through data to identify the key relationships and levers that drive sales.GenAI becomes even more powerful when it is combined with the expert forecasting capabilities of AI/ML,which together can provide a view of the past as well as a possible future.These goals can then be executed by AI

110、agents(discussed in more detail in the next section of this report).While GenAI can work with unstructured data,such as text,audio and video,it really shines when working with clean,well-organized data,such as data contained in a data lakehouse,which organizes the storage of both structured and unst

111、ructured data in a format that is ideal for AI(see Figure 5).DiscussionWhat to Know NowOften,enterprise data is siloed and unstructured.Whats Important in 2025GenAI enables everyone in an enterprise to access data stranded in siloes and in unstructured formats via natural language.Source:Coresight R

112、esearchFigure 5.Timeline of Evolution of Data RepositoriesOperations Read More on This Trend:AI and Unified Data:Empowering Next-Generation Product and Shopper Intelligence14OperationsAgentic AI Is Here82025 will be the year of the rise of AI agents,and their use will likely flourish in 2025 as user

113、s begin to build their own agents.AI agents represent the third wave in the commonly espoused view of the waves of AI development.The first wave is predictive,which is provided by AI/ML.We are currently on the border of the secondgenerative,in which AI can create media,synthetic data and computer co

114、deand third waves.In the third wave,AI will be able to power agents that operate on our behalf and even communicate with other agents.In the speculative fourth wave,computers would gain human-like powers of intelligence and reasoning(see Figure 6).Fortunately,the new generation of agents does not re

115、quire a step-up in computing power:AI PCs equipped with AI acceleration are able to operate agents effectively,and they can greatly benefit the in-store applications mentioned earlier in this report.Salesforce was one of the first enterprise software companies to anticipate and announce AI agents.Th

116、e company amplified the agent buzz by renaming its annual September“Dreamforce”conference as“Agentforce,”reinforcing the importance of agents.Agentforce is also the name of Salesforces platform that embodies its vision of AI agents:In addition to providing prebuilt agents for the companys sales,mark

117、eting,commerce and other cloud platforms,Agentforce also includes tools for users to create their own agents.Other cloud software vendors that have made their own agent announcements include Google Cloud,Snowflake and Workday,among others.There is a great deal of utility to be realized from agents,f

118、irst on the enterprise level and later on the consumer level,as agents automatically handle routine business tasks and personal tasks.Read More on This Trend:The Future of AI in Retail:What Technology Could Make Possible by 2030DiscussionSource:Salesforce/Coresight ResearchFigure 6.Waves of AI Devel

119、opmentWave 1PredictiveWave 2GenerativeWave 3AutonomousWave 4Advanced General IntelligenceWhat to Know NowSeveral major enterprise software providers have announced the availability of AI agents,as well as the tools required for creating them.Whats Important in 2025AI agents are likely to see mainstr

120、eam deployment in 2025 and beyond.15OperationsGenAI Can Unlock Supply Chains9Supply chains contain a heterogeneous collection of manufacturers,intermediaries,service providers,and a potentially diverse number of products,which are often produced across the globetogether,these factors make data colle

121、ction and communication a significant challenge.The concept of an interconnected supply chainone running on a base of shared data,as depicted in Figure 7 on the rightremains an ideal worth pursuing;however,in the meantime,the world remains a chaotic place.Still,retailers and brands require data on a

122、n items status and location from disparate nodes across the supply chain,in as near to real time as possible,in order to make important business decisions that affect inventory assortment and allocation,which are key to customer sentiment.There are multiple powerful supply chain management and analy

123、sis platforms,which have typically been the product of writing and perfecting computer code and designing dashboards;however,consumer preferences change by the minute,and the world is an unpredictable,dynamic place,constantly reducing the accuracy and relevance of these traditional methods.As mentio

124、ned in other sections of this report,GenAI enables the ad hoc analysis of structured and unstructured data,which is particularly important in regard to supply chains as data from overseas suppliers is unlikely to be clean and formatted.AI/ML can triage emerging supply chain issues,automatically hand

125、ling non-urgent issues and alerting operators about issues that require human intervention.GenAIs analytical and generative abilities enable it to adapt to unplanned or emerging situations quickly without the need for additional programming.Read More on This Trend:All our coverage of supply chain an

126、d GenAI in retail.DiscussionSource:Coresight ResearchFigure 7.The Continuous,Interconnected Product-Management Cycle of the FutureWhat to Know NowGenAI can acquire unstructured data from the entire supply chain and manage unplanned supply chain events.Whats Important in 2025AI agents will increasing

127、ly handle non-urgent issues in supply chains in 2025 and beyond.16OperationsDemand Forecasting 3.010What to Know NowAdvanced demand forecasting tools leverage ML and external data to create highly accurate forecasts.Whats Important in 2025GenAI offers the ability to incorporate unstructured data to

128、improve forecasts and to query forecasts using natural language.Demand forecasting plays a central role in a retailers operationsgoing beyond just planningas the results from a demand forecast determine the function and limits of other operations,such as sales and inventory operations,as illustrated

129、 in Figure 8.There are also several levels of demand forecasting,depending on the specific applicationin Figure 9,we illustrate the levels of demand forecasting sophistication.ML-based forecasting is about 50%accurate,according to industry experts,and incorporating external data,such as the weather

130、and local calendars,can raise its accuracy to more than 90%;adding other external data,such as demand signals from social media,can increase the accuracy of demand forecasting even further.Certain demand forecasts require additional constraints and levels of optimization;for instance,when forecastin

131、g demand for fresh food in grocery,forecasts require constraints in order to manage spoilage and reduce food waste.GenAI Leverages unstructured data Natural-language interfaceAdvanced AI/ML Incorporates external data,such as weather or local calendars,for greater accuracyAutomation Structured,robust

132、 approach Ends reliance on spreadsheets and human instinct Limits human interventionDiscussionLevels of Demand ForecastingBenefitsSource:Coresight ResearchSource:Coresight ResearchFigure 8.Information Flows Linking Demand Forecasting,Operations and OutcomesFigure 9.Increasingly Advanced Levels of De

133、mand Forecasting with Expected BenefitsAs discussed in other sections of this report,the advent of GenAI offers innovative,new methods of obtaining and analyzing data,as well as presenting it,which extends to demand forecasting processes.Read More on This Trend:Retail-Tech Landscape:Demand Forecasti

134、ngIntermediateAdvancedBasic17OperationsWhat We ThinkRetailers and brands are currently in the enviable position of having several powerful technologies to choose from to make their businesses more productive and efficient,as well as to entrench themselves with their customers.AI appears practically

135、everywhere throughout our list of retail technology trends,from computer vision and visual shopping and search to data analysis and personalization.Building on this potential,GenAI promises to bring the power of AI to everyone,and,though in the early stages of development,AI agents will be able to a

136、ct on our behalf,bringing even more productivity gains and new capabilities to the masses.Retailers need to consider bringing some computing capabilities back to the premises,or the edge,to enable CV and many AI functions inside of their stores.The confluence of CV,voice recognition and AI will offe

137、r new opportunities for retailers and brands to market their products.GenAI enables personalization at scale,and savvy retailers and brands should look to capitalize on this.GenAI brings with it many new,flexible ways to analyze enterprise data and communicate its results.AI agents are an exciting n

138、ew technology that can offer new capabilities and productivity gains and can effectively run on cost-effective systems with integrated AI acceleration.Implications for Brands/Retailers18What We Think Walmarts size and scale as the worlds largest retailer has enabled it to demonstrate multiple succes

139、sful AI deployments,including an intelligent shopping assistant,a company knowledge base for employees and a cashier-free Sams Club that uses CV to verify purchases.Retailers that use AI-and CV-powered robots for inventory scanning and data collection,such as Simbe,experience fewer out-of-stock item

140、s and fewer price and promotion errors.Retailers that have supplied their associates with handheld mobile devices,including Office Depot and Walgreens,see improved inventory visibility,order accuracy and worker productivity,as well as enhanced customer experiences.Various retailers,including BJs Who

141、lesale Club,have increased revenue by using AI-powered pricing applications.Retailers and brands that continue to use Excel or human intuition for demand forecasting will see lower accuracy compared to those that use AI/ML-based forecasting tools.Additionally,companies that use forecasting tools wit

142、h GenAI interfaces will be able to quickly identify new insights and opportunities.Retailers who lag in equipping associates with their own mobile devices will lose out on the additional productivity and employee satisfaction they provide,reducing their employees ability to upsell and provide excell

143、ent customer service.Retailers and brands sending generic e-mail blasts to customers and not providing personalized webpages,communications or product recommendations will risk having their customers leave to companies that offer greater personalization.Retailers performing inventory and planogram c

144、hecks in an unautomated fashion risk stock-outs,a lack of planogram compliance and the loss of workers due to tedious assignments.Brands or Retailers Poised To Gain AdvantageBrands or Retailers That Risk Losing Advantage CV leverages computing power and AI to perform many powerful functions in the r

145、etail store,including functions centered around inventory,customer behavior and data collection.Enterprises can benefit from tools developed to process siloed and unstructured data,while GenAI enables employees to perform many functions previously only performed by data scientists.GenAI models and a

146、pplicationsas well as AI agentsare in their early stages,offering companies many new opportunities,business models and openings to leapfrog competitors.AI is a core technology enabling many retail operations,including demand forecasting,personalization,communication and CV,among many others.Many ret

147、ailers have successfully deployed GenAI via applications such as intelligent chatbots,company knowledge bases and product description and image creators,as well as various custom applications.AI agents and design tools are becoming increasingly available;these offerings can provide a new level of ca

148、pabilities and productivity enhancements to companies across the retail space.Impacts from AIImplications for Technology Vendors19What We ThinkNotesData in this report are as of January 2,2025.Companies mentioned in this report are:Advanced Micro Devices(NasdaqGS:AMD),Alphabet Inc.(parent company of

149、 Google;NasdaqGS:GOOGL),Amazon(NasdaqGS:AMZN),Apple(NasdaqGS:AAPL),Avery Dennison Corporation(NYSE:AVY),BJs Wholesale Club Holdings(NYSE:BJ),GK Software/Fujitsu Limited(TSE:6702),Grabango,Hewlett Packard Enterprise Company(NYSE:HPE),HP Inc.(NYSE:HPQ),Intel Corporation(NasdaqGS:INTC),Instacart/Mapleb

150、ear(NasdaqGS:CART),Kroger(NYSE:KR),Nvidia Corporation(NasdaqGS:NVDA),Rockbot,Samsung Electronics(KOSE:A005930),Shopic,Snowflake(NYSE:SNOW),SpartanNash Company(NasdaqGS:SPTN),Super Micro Computer(NasdaqGS:SMCI),The ODP Corporation(parent company of Office Depot;NasdaqGS:ODP),Trax Retail,Trigo,VusionG

151、roup(ENXTPA:VU),Walgreens Boots Alliance(NasdaqGS:WBA),Walmart(NYSE:WMT),Workday(NasdaqGS:WDAY),Zebra Technologies(NasdaqGS:ZBRA)20NotesAbout Sponsored ReportsAbout IntelSponsored reports from Coresight Research are made available to nonsubscribers through commercial partnerships with retail or tech

152、nology firms.These reports,which present expert analysis and proprietary data,are prepared by Coresight Research but may incorporate input into their content from the sponsoring company.This report is sponsored by Intel.Intel is an industry leader,creating world-changing technology that enables glob

153、al progress and enriches lives.Inspired by Moores Law,Intel continuously works to advance the design and manufacturing of semiconductors to help address its customers greatest challenges.By embedding intelligence in the cloud,network,edge and every kind of computing device,the company unleashes the potential of data to transform business and society for the better.Steven WinnickVice PresidentInnovator ServicesJohn Harmon,CFAAssociate Director of Technology Research To discuss our research or partnership opportunities,email New York Hong Kong Lagos London Mangaluru(India)S21Notes