利用人工智能代理推动商业价值：与Walter Sun博士的对话

Driving businessvaluewith AI agents A conversation withDr. Walter SunHead of AI, SAP Executive Conversations DR. WALTER SUNHead of AI SAP DRIVINGBUSINESS VALUEWITH AI AGENTS Dr. Walter Sun is Senior Vice President and GlobalHead of Artificial Intelligence at SAP. He leads acentralized AI unit that engineers AI products forimplementation and reuse across SAP applications.Prior to SAP, Dr. Sun was Vice President ofCopilot Applied Artificial Intelligence for businessapplications at Microsoft. Previously, Dr. Sunworked at BlackRock Financial Management as aquantitative portfolio analyst and at Apple Inc. as asenior software engineer and scientist. SAP’S AI STRATEGY: THE 3RS FRAMEWORK How is SAP integrating AI intobusiness operations? Dr. Walter Sun:SAP follows the “3Rs”framework to ensure AI solutions areimpactful and trustworthy: 1.Relevant –AI must be tailoredto meet business needs. Weincorporate industry-specific contextinto our AI models. For example, asupply chain company in the US mayhave different requirements thanone in the UK, and our AI adaptsaccordingly. Dr. Walter SunHead of AISAP 2.Reliable –AI must provide accurate,fact-based outputs. We usegrounding techniques [connectingmodel output to verifiable sources]and provenance checks to verify AI-generated information, minimizinghallucination and maximizingaccuracy. 3.Responsible –AI must be ethical,explainable, and compliant withregulations. Transparency is at thecore of our approach, ensuring usersunderstand the logic behind AI-driven decisions. WE USE GROUNDINGTECHNIQUES [CONNECTINGMODEL OUTPUT TOVERIFIABLE SOURCES]AND PROVENANCE CHECKSTO VERIFY AI-GENERATEDINFORMATION, MINIMIZINGHALLUCINATION ANDMAXIMIZING ACCURACY. SAP integrates AI into three key areas: •Native AI integration:AI is embedded into SAP applications likeSuccessFactors for HR management, allowing users to generate jobdescriptions using natural language. •Joule AI copilot:Joule is SAP’s digitalassistant, enabling users to execute tasksseamlessly across SAP applications. AI must be ethical,explainable, andcompliant withregulations" •Business Technology Platform (BTP):Ourgenerative AI (Gen AI) hub offers accessto over 30 large language models (LLMs),allowing businesses to build and customizeAI applications. How does Joule AI – SAP’s digital assistant – work? Dr. Walter Sun:Joule is SAP’s digital copilot, designed to facilitate naturallanguage interactions across business applications. It serves as a bridgebetween users and various AI agents, orchestrating tasks efficiently. For instance, if a user wants to schedule a business trip, Joule can: 1.Check both the user’s calendar and those of their colleagues to find aperiod when all relevant parties are available.2.Coordinate with SAP’s travel platform to book flights and hotels.3.Connect with CRM [customer relationship management] systems to notifyrelevant stakeholders and log the meeting. Joule stands out from other AI agents because of its deep integrationinto the SAP ecosystem. It can interact across multiple SAP applications– finance, supply chain, HR, customer relations – offering a unified, AI-driven experience. Unlike standalone AI agents, Joule connects variousenterprise functions, allowing organizations to streamline processes acrossdepartments. AI ADOPTION: DRIVERS AND CHALLENGES What are the key drivers behind businesses adopting AI andAI agents? Dr. Walter Sun:JThe primary motivation for businesses to adopt AI is todrive efficiency and create value. Organizations want to achieve more inless time, enhance their decision-making, and automate repetitive tasks. AItechnology – including AI agents – offers powerful tools to help businessesmeet these objectives. Another important factor is customer demand. End-consumers andenterprise clients expect seamless, intelligent interactions. AI can enhancecustomer service, personalize user experiences, and improve overall businessefficiency. But AI adoption is not just aboutautomation. It is about stayingcompetitive. Historically, every majortechnological shift has created winnersand laggards. Organizations thatrecognized the advantages of PCs earlyon gained an advantage over those thatstuck with typewriters. Similarly, whenthe internet emerged, businesses thatembraced online connectivity becamemore efficient than those that didn’t. AI adoption isnot just aboutautomation. Itis about stayingcompetitive" AI represents another paradigm shift.Companies recognize they must integrateAI into their operations to remaincompetitive. They want to explore AI’spotential while minimizing risks related tosecurity and privacy. What are some of the biggestchallenges when adopting AI? Dr. Walter Sun:Many organizationsare eager to explore AI but are still inthe early stages and are proceedingcautiously. One of their biggestconcerns is around security and privacy.Organizations want to harness AI whileprotecting their data. They also wanttransparency on how AI models makedecisions,