行业研究公司研究宏观策略财报招股书会议纪要海南封关低空经济 DeepSeek AIGC 大模型

构建、购买还是混合？在组织中部署人工智能

2025-03-31GartnerM***

AI智能总结

Every CIO and tech leader is facing the challenge of integrating AI into their organizations, especially with AI becoming a standard feature in many solutions. The Gartner AI Technology Sandwich offers a new framework to guide decisions on acquiring and deploying AI capabilities. Here are the key questions to consider:

How to Use the AI Technology Sandwich

As a communication tool: Helps governance teams, IT members, and senior business leaders understand how AI pieces fit together, including technology and governance elements.
As a planning tool: Serves as a conceptual framework for creating or testing technology stacks, planning AI decentralization, and financial planning for AI execution.

AI Applications

Embedded AI: Determine if AI will come from vendors (e.g., SAP, ServiceNow, Salesforce) and how the AI platform needs to evolve to support it.
Built AI: Assess the need for customized AI solutions and decide whether to build them internally or assemble them from purchased components.
Bring-your-own AI: Evaluate if departments will buy AI independently and whether to develop governance frameworks and procure compliance tools.

Data Management

Data everywhere: Decide if integrating structured and unstructured data sources is necessary and how to procure data management solutions (build or buy).
Centralized data: Determine if data accessibility and compliance are required and whether to develop custom solutions or purchase existing tools.

Trust, Risk, and Security Management (TRiSM)

Additional security measures: Assess the need for security measures to ensure AI trustworthiness and decide whether to build them in-house or buy TRiSM technologies.
TRiSM oversight: Evaluate if real-time monitoring of AI outputs is critical and whether to develop custom oversight solutions or purchase established technologies.

Budget and Resources

Financial readiness: Determine if there is sufficient budget and resources to support AI development, implementation, and usage, and whether to use FinOps tools to monitor costs.

Strategic Alignment

Competitive goals: Assess if the AI buy vs. build strategy meets competitive goals and whether strategic vendor partnerships have all necessary players to support the AI strategy.

The AI Technology Sandwich helps organizations align their AI strategies with broader objectives, ensuring effective integration and governance of AI capabilities.

10 Questions to Ask AboutYour AI Build vs. Buy DecisionsGuiding your build vs. buy decisions using theGartner AI Tech Sandwich Every CIO and tech leader is grappling with how to effectivelyintegrate AI into their organizations, especially as AI becomes astandard feature in so many solutions.It’s time for CIOs to reimagine how their technology stacks,data approaches and governance structures will evolve tosupport AI. The Gartner AI Technology Sandwich offers a newprism through which to view your AI tech stack.Start with these 10 essential questions to guide your decisionson acquiring and deploying key AI capabilities. Gartner clients are using this checklist with their Executive Partners to build and execute their AI strategy.If you’re interested in learning how you can tap into the power of these former CIOs, click here. How to use the AI technology sandwich As a communication toolThe tech sandwich is a good way to show howall the AI “pieces” fit together — all on a singleslide. The tech sandwich helps:•The governance teams understand what’scoming and what they will need to govern.•Members of the IT organization get abetter sense of how AI will be present intheir areas.•Senior business leaders better understandthe technology and governance elementsrequired to execute AI. As a planning toolThe tech sandwich is not a tech stack ora reference architecture. It is a conceptualframework that can lead to the creation ortesting of many plans. The tech sandwich canbe used as:•An initial concept to start, test or finalizea formal technology stack or referencearchitecture.•To plan how the decentralization (ordemocratization) of AI should take placeso that the centralized IT and data scienceteams can support this better.•To plan financially for costs and resourcesthat will be required to execute AI inthe enterprise. AI applicationsFirst, consider where your AI capabilities will be spread. 1Embedded AIWill AI come from the AI that will beembedded by vendors in your purchasedapplications (SAP, ServiceNow, Salesforce)?If yes, how does your AI platform need to evolve to supportembedded AI? 2Built AIIs there a need for highly customized AIsolutions within your organization?If yes, will you build these solutions internally from scratch,or will you buy components and assemble from those? 3Bring-your-own AIWill your departments and business unitsbuy AI on their own?If yes, will you develop governance frameworks and procureexternal compliance tools? Data managementSecond, ask what data will be required coming from allAI applications. 4Data everywhereIs integrating both structured andunstructured data sources necessary foryour organization?If yes, how will you procure the necessary data managementsolutions? Build in-house or buy from vendors? 5Centralized dataDo you need to ensure data accessibilityand compliance for privacy andsafety reasons?If yes, will you develop custom solutions to manage dataquality, or will you purchase existing tools? Trust, risk and securitymanagement (TRiSM)Next, how will you make AI safe in your enterprise, no matterwhere it came from? Do you need additional security measuresto ensure the trustworthiness of yourAI deployments?If yes, will you build additional security measures in-house,or will you buy TRiSM technologies? 7TRiSMIs real-time monitoring of AI outputscritical for your organization, or can yourely on human-in-the-loop verification?If yes, will you develop custom oversight solutions, or will youpurchase established technologies? Budget and resourcesWill you be able to foot your AI bill? Do you have the budget and resources tosupport AI development/implementationand usage?If yes, have you properly costed out buy vs. build vs. use, andhave you purchased FinOps tools to monitor this? Strategic alignmentAsk yourself if this aligns to broader strategic objectives. 9Does your AI buy vs. build strategy meetyour competitive goals?Do you have enough build to maintain a competitive edge,or do you think you can differentiate enough with just a“buy” strategy? Does your strategic vendor partnerecosystem have all the players you needto support your AI strategy? Talk to one of our specialists today to learn moreabout how we can help you achieve your goals.Let’s Connect Build successwith personalizedadvice Actionable, objective insightExplore these additional complimentary resources and tools for CIOs:

点击免费查看完整报告