德勤：2024年人工智能代理如何重塑未來工作（英文版）（18頁）.pdf

《德勤：2024年人工智能代理如何重塑未來工作（英文版）（18頁）.pdf》由會員分享，可在線閱讀，更多相關《德勤：2024年人工智能代理如何重塑未來工作（英文版）（18頁）.pdf（18頁珍藏版）》請在三個皮匠報告上搜索。

1、Prompting for action How AI agents are reshaping the future of workExpanded capabilities,use cases and enterprise impact from Generative AINovember 2024 Deloitte AI InstituteAbout the Deloitte AI InstituteThe Deloitte AI InstituteTM helps organizations connect the different dimensions of a robust,hi

2、ghly dynamic and rapidly evolving AI ecosystem.The Institute leads conversations on applied AI innovation across industries,with cutting-edge insights,to promote human-machine collaboration in the“Age of With.”The Deloitte AI Institute aims to promote a dialogue and development of artificial intelli

3、gence,stimulate innovation,and examine both challenges to AI implementation and ways to address them.The Institute collaborates with an ecosystem composed of academic research groups,startups,entrepreneurs,innovators,mature AI product leaders and AI visionaries to explore key areas of artificial int

4、elligence including risks,policies,ethics,future of work and talent,and applied AI use cases.Combined with Deloittes deep knowledge and experience in artificial intelligence applications,the Institute helps make sense of this complex ecosystem,and as a result delivers impactful perspectives to help

5、organizations succeed by making informed AI decisions.No matter what stage of the AI journey youre in,whether youre a board member or a C-suite leader driving strategy for your organization or a hands-on data scientist bringing an AI strategy to life,the Institute can help you learn more about how o

6、rganizations across the world are leveraging AI for a competitive advantage.Visit us at the Deloitte AI Institute to access the full body of our work,subscribe to our podcasts and newsletter,and join us at our meetups and live events.Lets explore the future of AI for action|How AI agents are reshapi

7、ng the future of work2ContentKey takeaways AI agents are reshaping industries by expanding the potential applications of Generative AI(GenAI)and typical language models.Multiagent AI systems can significantly enhance the quality of outputs and complexity of work performed by single AI agents.Forward

8、-thinking businesses and governments are already implementing AI agents and multiagent AI systems across a range of use cases.Executive leaders should make moves now to prepare for and embrace this next era of intelligent organizational transformation.Introduction 4AI agents:5What makes them differe

9、ntand why they matter Multiagent AI systems:7Amplifying the potential of AI agents Key benefits of AI agents and multiagent AI systems:7Advantages that AI agents are unlocking for organizations today Transforming strategic insights:8A real-world example of a multiagent AI system Achieving impact thr

10、ough targeted use cases:11How AI agents are changing industries and enterprise domains Enabling new ways of working and new horizons of innovation:13Implications for strategy,risk,talent,business processes and technology The road ahead:15What we expect as AI agents continue to evolve Charting a cour

11、se into the next era of organizational transformation:16Recommended actions for leaders to take now Get in touch&Endnotes 17Prompting for action|How AI agents are reshaping the future of work3How can we operate faster and more efficiently?This question has always been at the forefront of strategic a

12、gendasbut Generative AI(GenAI)is helping unlock new answers.With its ability to produce novel outputs from plain-language prompts,GenAI has enabled enterprises to significantly enhance speed and productivity across a range of business tasks.However,use cases for typical language models have only jus

13、t begun to show GenAIs transformative potential.In this time of rapid AI evolution,its time to think bigger and bolder:from streamlining routine tasks to redesigning entire workflows.Now the question for business and government leaders is becoming:How can we rethink our business processes with GenAI

14、?Large language models(LLMs)and GenAI-powered tools used by most organizations today serve as helpful assistants:A human worker enters a prompt,GenAI quickly produces an output.However,this interaction is largely transactional and limited in scope.What if GenAI could be more like a skilled collabora

15、tor that will not only respond to requests but also plan the whole process to help solve a complex need?What if GenAI could also tap into the necessary data,digital tools and contextual knowledge to orchestrate the process end to end,autonomously?This vision is becoming a reality with the emergence

16、of AI agents and multiagent AI systemsa powerful advancement in whats possible through human-AI partnership.Leading companies and government agencies are already seeing the value of AI agents and putting them into practice.In this paper,we explore what makes AI agents so groundbreaking.We then revea

17、l how they are reshaping industries,including government and public services,by enabling new use cases,enhancing automation and accelerating the future of intelligent organizational transformation.IntroductionAdapt or fall behindAt the end of 2023,nearly 1 in 6 surveyed business leaders said GenAI h

18、ad already transformed their businesses.1Prompting for action|How AI agents are reshaping the future of work4AI agents:What makes them differentand why they matter To grasp the potential value of AI agents and their role in expanding the automation horizon,it is important to understand how they diff

19、er from the language models and GenAI applications familiar to business leaders today.As a result,early GenAI use cases have mostly been limited to standalone applications such as generating personalized ads based on a customers search history,reviewing contracts and legal documents to identify pote

20、ntial regulatory concerns,or predicting molecular behavior and drug interactions in pharmaceutical research.AI agents excel in addressing these limitations while also leveraging capabilities of domain-and task-specific digital tools to complete more complicated tasks effectively.For example,AI agent

21、s equipped with long-term memory can remember customer and constituent interactionsincluding emails,chat sessions and phone callsacross digital channels,continuously learning and adjusting personalized recommendations.This contrasts with typical LLMs and SLMs,which are often limited to session-speci

22、fic information.Moreover,AI agents can automate end-to-end processes,particularly those requiring sophisticated reasoning,planning and execution.AI agents are opening new possibilities to drive enterprise productivity and program delivery through business process automation.Use cases that were once

23、thought too complicated for GenAI can now be enabled at scalesecurely and efficiently.In other words:AI agents dont just interact.They more effectively reason and act on behalf of the user.AI agents are reasoning engines that can understand context,plan workflows,connect to external tools and data,a

24、nd execute actions to achieve a defined goal.While this may sound broadly like what standalone LLMs or GenAI applications can do,there are key distinctions that make AI agents significantly more powerful.(See table,page 6.)Typical LLM-powered chatbots,for example,usually have limited ability to unde

25、rstand multistep promptsmuch less to plan and execute whole workflows from a single prompt.In essence,they conform to the“input-output”paradigm of traditional applications and can get confused when presented with a request that must be deconstructed into multiple smaller tasks.They also struggle to

26、reason over sequences,such as compositional tasks that require consideration of temporal and textual contexts.These limitations are even more pronounced when using small language models(SLMs),which,because they are trained on smaller volumes of data,typically sacrifice depth of knowledge and/or qual

27、ity of outputs in favor of improved computational cost and speed.Prompting for action|How AI agents are reshaping the future of work5Typical language modelsAI agentsAutomate tasksAutomate entire workflows/processesAre not capable of planning or orchestrating workflowsCreate and execute multistep pla

28、ns to achieve a users goal,adjusting actions based on real-time feedbackDo not retain memory and have limited fine-tuning capabilitiesUtilize short-term and long-term memory to learn from previous user interactions and provide personalized responses;Memory may be shared across multiple agents in a s

29、ystemAre not inherently designed to integrate with external tools or systemsAugment inherent language model capabilities with APIs and tools(e.g.,data extractors,image selectors,search APIs)to perform tasksRely on static knowledge with fixed training cutoff datesAdjust dynamically to new information

30、 and real-time knowledge sourcesTypically lack self-assessment capabilities and are limited to probabilistic reasoning based on training dataCan leverage task-specific capabilities,knowledge and memory to validate and improve their own outputs and those of other agents in a systemUse case scopePlann

31、ingMemory&fine-tuningTool integrationData integrationAccuracyA new paradigm for human-machine collaboration Through their ability to reason,plan,remember and act,AI agents address key limitations of typical language models.Prompting for action|How AI agents are reshaping the future of work6While ind

32、ividual AI agents can offer valuable enhancements,the truly transformative power of AI agents comes when they work together with other agents.Such multiagent systems leverage specialized roles,enabling organizations to automate and optimize processes that individual agents might struggle to handle a

33、lone.Multiagent AI systems:Amplifying the potential of AI agents CapabilityAI agents can automate interactions with multiple tools to perform tasks that standalone language models were not designed to achieve(e.g.,browsing a website,quantitative calculations).ProductivityWhereas standalone LLMs requ

34、ire constant human input and interaction to achieve desired outcomes,AI agents can plan and collaborate to execute complex workflows based on a single promptsignificantly speeding the path to delivery.Self-learningBy tapping short-and long-term contextual memory resources that are often unavailable

35、in a pre-trained language model,AI agents can rapidly improve their output quality over time.AdaptabilityAs needs change,AI agents can reason and plan new approaches,rapidly reference new and real-time data sources,and engage with other agents to coordinate and execute outputs.AccuracyA key advantag

36、e of multiagent AI systems is the ability to employ“validator”agents that interact with“creator”agents to test and improve quality and reliability as part of an automated workflow.IntelligenceWhen agents specializing in specific tasks work togethereach applying its own memory while utilizing its own

37、 tools and reasoning capabilitiesnew levels of machine-powered intelligence are made possible.TransparencyMultiagent AI systems enhance the ability to explain AI outputs by showcasing how agents communicate and reason together,providing a clearer view of the collective decision-making and consensus-

38、building process.Key benefits of AI agents and multiagent AI systemsMultiagent AI systems employ multiple,role-specific AI agents to understand requests,plan workflows,coordinate role-specific agents,streamline actions,collaborate with humans and validate outputs.Multiagent AI systems typically invo

39、lve standard-task agents(e.g.,user interface and data management agents)working with specialized-skill and-tool agents(e.g.,data extractor or image interpreter agents)to achieve a goal specified by a user.At the core of every AI agent is a language model that provides a semantic understanding of lan

40、guage and contextbut depending on the use case,the same or different language models may be used by agents in a system.This approach can allow some agents to share knowledge while others validate outputs across the systemimproving quality and consistency in the process.That potential is further enha

41、nced by providing agents with shared short-and long-term memory resources that reduce the need for human prompting in the planning,validation and iteration stages of a given project or use case.This concept extends whats possible with individual AI agents by taking a team or agency approach.By decom

42、posing a detailed process into multiple tasks,assigning tasks to agents optimized to perform the tasks,and orchestrating agent and human collaboration at each stage of the workflow,this type of system has proven much more likely to produce higher quality,faster and more trustworthy outcomes.2,3 In o

43、ther words:Multiagent AI systems dont just reason and act on behalf of the user.They can orchestrate complex workflows in a matter of minutes.Prompting for action|How AI agents are reshaping the future of work7No matter the industry,every organization engages in research,analysis and reportingwhethe

44、r about economic conditions,customer and constituent preferences,policy and pricing strategies,or other topics.Traditionally,these projects require skilled human analysts to perform multiple steps,which can be time-consuming,utilizing research and analysis tools along with in-house subject matter ex

45、pertise.Transforming strategic insightsWhile effective and repeatable,this approach is Time-consuming Completing a single report can take days or weeks,making it difficult to seize emerging opportunities.Inefficient Skilled analysts must perform many repetitive activities that take their focus away

46、from higher-level analysis.Difficult to scale Companies and government agencies can struggle to hire and retain enough skilled,experienced analysts to grow their research capacity.Heres what a traditional research project typically looks like.AnalystAnalystAnalystStakeholderAnalystStakeholderAnalyst

47、 or DesignerProoferRisk&complianceAnalyst identifies topic and scope:A report on the top 5 GenAI trends in financial services,based on publicly available data from the prior 3 months.Stakeholder provides feedback on outline.Analyst selects sources,searches and compiles relevant information,and organ

48、izes materials and notes.Analyst drafts the report and sends to stakeholder,who provides feedback and iterates with analyst.Analyst sends approved report to designer.Analyst or designer researches images,develops graphics and designs report.Proofer reviews report and provides feedback,which analyst

49、and/or designer incorporate.Risk&compliance professionals are engaged as needed.Analyst synthesizes themes and perspectives,outlines a plan for the report and sends to business stakeholder for review.Final report is delivered.Prompting for action|How AI agents are reshaping the future of work8Deloit

50、te has developed a multiagent AI system that can streamline and improve each step of research and reporting.Heres how it works.In addition to being effective and repeatable,this AI agent-powered approach is Fast A single,quality report can be produced in less than an hour.Efficient Skilled professio

51、nals can focus on validating,iterating and refining the report.Highly scalable In essence,this system provides an instantly available team of skilled digital workers.Planning agent breaks the goal into subprocesses,develops a workflow and identifies necessary tools and specialized agents to execute

52、the workflow.Analyst and interface agent discuss and define report scope,sources and timeframe for data collection,target industry and audience,etc.Through this process,the analyst defines the deliverable:A report on the top 5 GenAI trends in financial services,based on publicly available data from

53、the prior 3 months.Specialized agents expand prompts,conduct research,compile and analyze results,identify themes and draft the report outline.As needed,the multimodal processing agent translates and interprets data collected from visual and audio sources.Once the outline is approved/adjusted by the

54、 analyst,additional specialized agents draft and design the report complete with customized charts and illustrations.Throughout the process,the quality assurance agent checks for accuracy,quality and regulatory/brand compliance,while the data management agent ensures source materials and report iter

55、ations are documented for reference/review.Analyst reviews the report and requests changes.The system iterates and refines the report.AnalystUser interfaceFile managementMultimodal processingPlanning“I need to write a report about GenAI trends in my industry.”“Please tell me about your request.”Prom

56、pt expandingData sourcingWeb browsingTopic modelingContent summarizationReport formattingData structuring Data visualizingQuality assuranceImage selectionReport writingAnalystAI AGENT TYPESStandard-task agent(s)One or more agents that perform tasks common to all workflowsSpecialized-skill&-tool agen

57、tsRole-specific agents that execute specific tasks within the workflowFinal report is delivered.All agents can access Language models(shared or separate)External tools&data sources as needed Shared short-and long-term memory Prompting for action|How AI agents are reshaping the future of work9Effecti

58、ve and efficient work depends on creativity and knowledge augmented by well-planned processes and task-appropriate tools.Thats what AI agents and multiagent AI systems can bring together.Prompting for action|How AI agents are reshaping the future of work10 USE CASE Dynamic pricing and personalized p

59、romotionsINDUSTRY:ConsumerStandard pricing strategies often involve static models that do not account for real-time market conditions,customer behavior or inventory levels.Multiagent AI systems can rapidly integrate analysis based on vast amounts of real-time datasuch as competitor pricing,customer

60、purchase history and seasonal trendsto dynamically adjust prices.Additionally,they can personalize promotions based on individual customer preferences,attributes and shopping habits with the goal of improving conversion rates and elevating customer satisfaction.POTENTIAL ADVANTAGES ACHIEVED WITH AI

61、AGENTS:Achieving impact through targeted use cases Organizations across industries and sectors are already leveraging the potential of AI agents and multiagent systems to transform processes,improve efficiency,and expand impact.Lets explore four use cases that are possible todaytwo in specific indus

62、tries,and two that can be applied in any business.USE CASE Individualized financial advisory and wealth managementINDUSTRY:Financial servicesFinancial advisory services often have relied on broad categorizations of customers based on age,income and risk tolerance.This approach can often miss the com

63、plexities of individual financial situations and goals.In todays rapidly changing financial landscape,there is an increasing demand for personalized,adaptive financial advice.Multiagent AI systems can analyze diverse data sourcesincluding the customers financial history,real-time market data,life ev

64、ents and even behavioral patternsto help advisers create financial plans and investment strategies tailored for the specific individual.AI agents can then continuously monitor and adjust recommendations as circumstances change.POTENTIAL ADVANTAGES ACHIEVED WITH AI AGENTS:Enhanced scalability Serve a

65、 larger number of customers with high-quality,personalized advice without raising costs to deliver.HyperpersonalizationCustomize financial advice to each customers specific needs and goals,considering factors that other methods might overlook.Continuous fine-tuning Automatically update financial pla

66、ns and strategies in response to changes in market conditions or personal circumstances.Improved customer satisfaction Strengthen customer relationships by providing more relevant and timely advice,leading to higher retention and satisfaction.Faster adaptation Adjust prices instantly in response to

67、market changes,inventory levels or customer demandoptimizing revenue.Greater profitability Maximize margins and minimize discounting by optimizing pricing and promotions on an ongoing basis.Personalized offers Tailor promotions to each customers preferences and behavior,increasing the likelihood of

68、purchase.12Prompting for action|How AI agents are reshaping the future of work11 USE CASE Talent acquisition and recruitmentDOMAIN:Human resources(HR)Traditional recruitment processes often involve manual resume screening,repetitive candidate assessments and significant administrative workwhich can

69、lead to inefficiencies.AI agents can automate the end-to-end recruitment process by using natural language processing to analyze resumes,assess candidates based on skills and experience,and conduct initial screening interviews via GenAI-powered avatars.These systems can collaborate with HR professio

70、nals to ensure that qualified candidates are identified,prioritized and moved through the hiring pipeline efficiently while adhering to relevant regulations.POTENTIAL ADVANTAGES ACHIEVED WITH AI AGENTS:Dynamic scalability Handle large volumes of applications,making it easier to manage hiring campaig

71、ns or recruit for multiple roles simultaneously.Increased efficiency Automate tasks to allow HR teams to focus on strategic activities,shortening the time to hire.Improved candidate matching Analyze a broader range of data points to help match candidates to roles more accurately,improving the qualit

72、y of hires.Reduced bias By standardizing candidate assessments and focusing on skills and experience,AI agents can help address unconscious bias in the recruitment process.USE CASEPersonalized customer supportDOMAIN:Customer and beneficiary serviceTraditional customer and beneficiary support systems

73、 often rely on scripted interactions,which can fail to resolve complex or unique inquiriesleading to customer frustration and escalation.In contrast,multiagent AI systems can understand plain-language requests and generate relevant and natural responses that consider the customers history,preference

74、s and real-time context.These advanced systems can handle many complex inquiries effectivelyreducing the need for escalation to live agents while improving customer/beneficiary satisfaction.POTENTIAL ADVANTAGES ACHIEVED WITH AI AGENTS:Greater consistency and scalability AI agents can operate 24/7 wi

75、thout fatigue,maintaining a consistent quality of service no matter the volume of inquiries.Compounding efficiencies The ability to learn from each interaction can help reduce response times,improve quality,and free up human service agents to focus on more nuanced customer requests.Improved customer

76、 experiences Each customer interaction can be adjusted to individual needs,improving satisfaction and engagement.34Prompting for action|How AI agents are reshaping the future of work12Enabling new ways of working and new horizons of innovationAs language models continue to evolve,AI agents and syste

77、ms are likely to become strategic resources and efficiency drivers for core business and government activities such as product development,regulatory compliance,customer service,constituent engagement,organizational design and others.We see a future in which agents will transform foundational busine

78、ss models and entire industries,enabling new ways of working,operating and delivering value.Thats why its important for C-suite and public service leaders to begin preparing now for this next chapter in the evolution of human-machine collaboration and business innovation.Lets explore some of the new

79、 ways of thinking and leading that should be considered during this time of rapid change.Leaders should begin integrating AI agents and multiagent AI systems into their overall strategies and future road maps.This involves reimagining business processes,investing in AI capabilities,and fostering cul

80、tures of innovation.Organizations should develop their own clear road map for AI agent adoption,identifying key areas where they can drive the most value and impact on broader business goals.Effective change management will be crucial for successful integration.Leaders should think carefully through

81、 how they will address organizational resistance,provide training,and ensure that employees understand the value and benefits of AI agents.This includes developing a comprehensive communication strategy to keep employees and other stakeholders informed and engaged throughout the adoption process.Str

82、ategy implicationsFOCUS AREAS Identify and prioritize business and service areas where AI agents can have the most immediate and measurable impact.Develop robust training programs to help employees understand and use AI agents in ways that improve productivity and efficiency.AI agents introduce new

83、risks that necessitate robust security and governance structures.A significant risk is potential bias in AI algorithms and training data,which can lead to inequitable decisions.Additionally,AI agents can be vulnerable to data breaches and cyberattacks,compromising sensitive information and data inte

84、grity.The complexity of AI systems also presents the risk of unintended consequences due to AI agents behaving unpredictably or making decisions not aligned with organizational goals.To manage these risks,it is important to set clear parameters for agent interactions,monitor operational metrics,and

85、continually ensure data ethics,privacy,security and integrity.As AI agents are integrated into core business processes,an enterprisewide governance framework with guidelines on data usage,ethics and security can further help mitigate risks.This framework should ensure compliance with relevant regula

86、tions and include continuous monitoring of AI agent interactions.Advanced security measures,such as encryption and multifactor authentication,can help protect against data breaches and cyberattacks.Training and awareness programs for employees can provide an additional defense by helping employees u

87、nderstand the ethical and operational considerations of working with AI agents.Risk implicationsFOCUS AREAS Identify brand and operational risks that may arise around data usage,AI agent interactions with each other and with tools,and ethics.Ensure model outputs are effectively tested and validated.

88、Implement an AI agent governance framework that is regularly reviewed and updated as AI technologies evolve.Monitor emerging risks specific to AI agents such as“agent autonomy”i.e.,the risk of unintended consequences when agents make decisions with minimal human oversight.Prompting for action|How AI

89、 agents are reshaping the future of work13The implementation of AI agents is likely to change the traditional workforce structure.As AI agents take over routine and lower-value tasks,there will likely be a high demand for human skills related to designing,implementing and operating these systems.Lea

90、ders should think through what new roles,job descriptions and job architectures are involved in building out the capability and then how to identify,recruit,train and retain this specialized talent.Beyond the implications for tech talent,enterprise leaders should be ready to help employees across a

91、wide variety of roles learn how to work with AI agents and even identify new use cases where they could improve processes.Deployed and managed well,AI agents can open up new realms of potential for human-machine collaborationbut that potential depends on workers understanding,embracing and being abl

92、e to perform new roles.Talent implicationsFOCUS AREAS Communicate the benefits of AI agents,and help employees adapt to new ways of working.Foster a culture of innovation and continuous learning.Leaders should instill a mindset of innovation and adaptability related to AI agents.Explore a redesign o

93、f job architectures,workflows and performance metrics to reflect the new reality of humans and AI agents working in tandem.AI agents and multiagent AI systems demand careful human evaluation of business processessometimes from the ground up.While agents will redefine many core processes over time,AI

94、 agents can be integrated into existing operating models today,enhancing the efficiency of current processes without the need for complete system overhauls.This approach makes it easier for organizations to adopt lower-risk agent solutions incrementallybut requires careful planning,management and al

95、ignment to ensure that AI agents are improving what people and/or other technology solutions already do well.In use cases where AI agents do make sense to implement,human involvement will remain vitally important for tasks requiring judgment,validation and critical decision-making.This collaboration

96、 is important to help ensure that AI outputs are accurate,reliable and effective.In this paradigm,everyone working with AI agents serves as a managergiving orders(via prompts),clarifying requests,monitoring progress,reviewing outputs and requesting or making changes as necessary.Business process imp

97、licationsFOCUS AREAS Ensure that where agents are implemented into existing business processes,those processes remain effective while driving greater efficiency and value.Establish processes for continuously monitoring and improving the performance of AI agents.This includes collecting and analyzing

98、 data on the performance of AI agents,identifying opportunities for improvement,and making changes as needed to optimize their performance.Prompting for action|How AI agents are reshaping the future of work14Implementing AI agents can be costly,requiring substantial investment in technology and infr

99、astructure.Organizations should carefully evaluate the value proposition and return on investment;and develop a phased approach to use cases,with a focus on“low-hanging fruit”(i.e.,simpler use cases)that can lay the groundwork for more complex activations.Quality data is the foundation for AI agents

100、 to work effectively.If data is inaccurate,incomplete or inconsistent,the agents outputs and actions may be unreliable or incorrectcreating both adoption and risk issues.Its therefore essential to invest in robust data management and knowledge modeling.Adopting trustworthy AI practices is a key to m

101、itigating risks and ensuring ethical deployment.This includes developing AI agent solutions that are fair,transparent and accountable,and addressing potential biases in AI models.Technology and data implicationsFOCUS AREAS Put the right technology infrastructure in place to support the adoption and

102、implementation of AI agents(e.g.,AI orchestration platforms and scalable data lakes).Ensure data is properly organized,up to date and accessible to AI agents.This includes having well-defined data governance policies and procedures as well as continuous access to real-time data feeds to enable dynam

103、ic,accurate decisions.Establish processes for monitoring and managing the performance and ethics of AI agents and multiagent AI systems.Without transparent and trustworthy AI,customer trust and regulatory compliance are at risk.The era of AI agent collaboration is still in its early stages.Interest

104、is growing among businesses and technology providers,but comprehensive solutions are not yet common.There is much technical work to be doneparticularly in terms of the reasoning and planning capabilities that will enable AI agents.Improvements are likely to come fast.In recent months GenAI tools hav

105、e shown significant improvements in reasoning and agent orchestration capabilities.Many venture capital firms are investing heavily across the spectrum of AI agent-related technologies,as are many of todays leading GenAI and technology providers.What is available today is only a glimpse of whats to

106、come.Indeed,we anticipate a significant evolution of core language models,AI agents,and agent orchestration platforms within the next 12 months.Future-focused leaders arent waiting on the sidelines.Across industries,companies are already designing,testing andin some casesimplementing agents.The road

107、 aheadPrompting for action|How AI agents are reshaping the future of work15Charting a course into the next era of organizational transformation AI agents and multiagent AI systems represent more than just technological advancements.They represent a fundamental shift in how organizations can automate

108、 processes,improve human-machine collaboration,generate insights and respond dynamically to complex challenges.They offer the potential to unlock significant value across a wide range of functionsfrom enhancing customer interactions and optimizing supply chains to driving innovation in product devel

109、opment and service delivery.The journey to realizing these benefits requires deliberate planning,strategic investments,and a commitment to fostering a culture that embraces continuous improvement and technological advancement.By aligning AI agent initiatives with core business goals,investing in the

110、 right infrastructure and nurturing a culture of innovation,your organization can be well-positioned to lead in this new era of AI-powered business transformation.Now is the time to move.GenAI tools are evolving rapidlyand that evolution is unlikely to slow down in the next few years.Similarly,AI ag

111、ents are already being implemented by companies across industries as well as by major technology providers.So,its important to begin exploring initial applications/use cases of agents,while setting the stage for future foundational business transformation.To begin your own organizations journey,cons

112、ider these actions:Assess and prioritize use casesBegin with a comprehensive assessment of your current operations to identify high-impact areas where AI agents can add value.Focus on processes that are ripe for automation,involve complex decision-making and/or require rapid adaptability.Prioritize

113、these use cases to achieve quick wins and demonstrate tangible value.Develop a strategic AI agent road mapAlign your AI initiatives with broader business and mission objectives by creating a detailed road map that outlines the integration of AI agents into your operations.This plan should include cl

114、ear milestones,timelines and success metrics to guide the deployment of AI agent-powered capabilities across the organization.Invest in infrastructure and human talent developmentIdentify and build the necessary infrastructure to support AI agents,including scalable cloud platforms,advanced data ana

115、lytics tools and robust cybersecurity measures.Simultaneously,invest in upskilling your workforce,focusing on technical skills and the ability to collaborate effectively with AI agents and multiagent systems.A well-prepared workforce is key to realizing the full transformation potential of AI agents

116、.Implement strong data governance and risk managementAs AI agents become integral to your operations,its important to establish strong governance frameworks to manage the associated risks.Implement policies that ensure data integrity,security and ethical use,while continuously monitoring AI interact

117、ions to safeguard against biases and unintended consequences.And compliance with regulatory standards should always be a top priority.Nurture a culture of innovationExperimentation and continuous learning are vital to your success.Empower your teams to explore new applications of GenAI,iterating on

118、initial deployments to drive ongoing improvements.By embedding innovation into the fabric of your organization,you can maintain a competitive edge in a rapidly changing business environment.12345Prompting for action|How AI agents are reshaping the future of work16Get in touchVivek Kulkarni Managing

119、Director,AI Transformation Deloitte LLP Prakul Sharma Principal,AI&Data Deloitte Consulting LLP Ed Van Buren Principal,GPS Applied AI Leader Deloitte Consulting LLP Endnotes1.Deborshi Dutt,Beena Ammanath,Costi Perricos and Brenna Sniderman,Now decides next:Insights from the leading edge of generativ

120、e AI adoption,Deloitte,January 2024,p.8,https:/ September 16,2024.2.KaShun Shum,Shizhe Diao and Tong Zhang,Automatic Prompt Augmentation and Selection with Chain-of-Thought from Labeled Data,Cornell University,February 27,2024,https:/arxiv.org/abs/2302.12822,accessed September 16,2024.3.Boshi Wang,S

121、ewon Min,Xiang Deng,Jiaming Shen,You Wu,Luke Zettlemoyer and Huan Sun,Towards Understanding Chain-of-Thought Prompting:An Empirical Study of What Matters,Cornell University,June 1,2023,https:/arxiv.org/pdf/2212.10001,accessed September 16,2024.Scott Holcomb Principal,GenAI Transformation Leader Delo

122、itte Consulting LLP Caroline Ritter Manager,AI Transformation Deloitte LLP Contributors to this report:Jim Rowan,Parth Patwari,Rajib Deb,Brijraj Limbad,Hye Ra MoonPrompting for action:A series on AI agents and multiagent AI systems Learn key insights to help guide your organizations agent-enabled jo

123、urney of transformation.Prompting for action|How AI agents are reshaping the future of work17About DeloitteAs used in this document,“Deloitte”means Deloitte Consulting LLP,a subsidiary of Deloitte LLP.Please see for a detailed description of our legal structure.Certain services may not be available

124、to attest clients under the rules and regulations of public accounting.Deloitte provides industry-leading audit and assurance,tax and related services,consulting,financial advisory,and risk advisory services to nearly 90%of the Fortune Global 500 and thousands of private companies.Our people deliver

125、 measurable and lasting results that help reinforce public trust in capital markets,enable clients to transform and thrive,and lead the way toward a stronger economy,a more equitable society,and a sustainable world.Building on its 175-plus year history,Deloitte spans more than 150 countries and terr

126、itories.Learn how Deloittes approximately 457,000 people worldwide make an impact that matters at .This publication contains general information only and Deloitte is not,by means of this publication,rendering accounting,business,financial,investment,legal,tax,or other professional advice or services

127、.This publication is not a substitute for such professional advice or services,nor should it be used as a basis for any decision or action that may affect your business.Before making any decision or taking any action that may affect your business,you should consult a qualified professional advisor.Deloitte shall not be responsible for any loss sustained by any person who relies on this publication.Copyright 2024 Deloitte Development LLC.All rights reserved.