DBS's Invisible Banking, Digital Transformation, Customer Journeys, and the Agentic Shift
DBS 的隱形銀行、數位轉型、顧客旅程與代理化轉移
Scroll to move through the session. Click, flip, sort, and inspect the animated workflow as you go.
向下捲動進入課程。你可以一路點擊、翻卡、排序，並查看動畫化工作流程。
Core idea: in the digital age, a brand's marketing promise—seamless, invisible, fast—must be supported by backend operating architecture. If the system is over-deconstructed in pursuit of customer experience, instability can directly destroy brand assets.
核心觀念：在數位時代，品牌的行銷承諾（如：無縫、隱形、快速）必須由後端營運系統的架構來支撐。當企業為了追求極致的顧客體驗而過度解構系統時，產生的不穩定性可能直接摧毀品牌資產。
This section asks you to break the one-way myth that digital transformation equals technology adoption, understand that excellent experience comes from end-to-end process reconfiguration, and use Reshuffle to preview AI-agent coordination.
這一部分要打破「數位轉型＝導入科技」的單向迷思，理解卓越體驗來自端到端流程重組，並透過《Reshuffle》預視 AI Agent 的新協作機制。
In this lab, you will use deep research tools, identify hallucinations and logic gaps, diagnose a real firm through Rewired and Reshuffle, and draw a first node-based service workflow for Session 2.
本部分將帶你使用深度研究工具、辨識 AI 幻覺與邏輯縫隙，利用《Rewired》與《Reshuffle》診斷真實企業，並為 Session 2 繪製第一張節點式工作流程圖。
Explore how DBS redesigns product and marketing communication from the customer's point of view.
探討 DBS 如何從客戶視角出發，重新設計產品與行銷溝通策略。
Background: DBS used to be mocked as “Damn Bloody Slow.” To change this, it proposed the vision “Make Banking Joyful” and changed its brand line to “Live More, Bank Less.” DBS positioned itself as a technology company and promoted the concept of “Invisible Banking,” hoping to embed financial services seamlessly into customers’ daily lives.
背景：過去 DBS 被戲稱為「Damn Bloody Slow」。為了改變這點，DBS 推出了「Make Banking Joyful」的願景，並將品牌標語改為「Live More, Bank Less」。DBS 將自己視為一家科技公司，推動「隱形銀行」的概念，希望將金融服務無縫嵌入顧客的日常生活中。
What is the brand promise?
這個品牌承諾要傳遞什麼？
Front side = the marketing statement
正面 = 行銷說法
DBS redefined the customer’s job-to-be-done. Customers do not go to a bank for the sake of “going to a bank”; they go because of a house, a car, and daily life. The focus of marketing communication therefore shifts from selling financial products to removing friction so that customers do not feel the bank’s presence.
DBS 重新定義了顧客的「任務（Job-to-be-done）」。顧客不是為了「來銀行」而來銀行，是為了買房、買車、生活。因此行銷溝通的重點不再是推銷金融產品，而是「消除摩擦力」，讓顧客感受不到銀行的存在。
Open the card to reveal the organizational and ecosystem changes.
翻開後可看見組織與生態系的變化。
DBS adopted an organizational redesign based on Managing through Journeys (MtJs) and used APIs extensively to build ecosystems—for example, working with Go-Jek or housing platforms—so that customers could complete financial transactions directly in other life scenarios.
導入「以顧客旅程為導向（Managing through Journeys, MtJs）」的組織重組，並大量使用 API 建立生態系（如與 Go-Jek 或房產平台合作），讓顧客能在其他場景直接完成金融交易。
Q: What is “Bank Less”? From the perspective of digital communication, what kind of value proposition is DBS trying to communicate to consumers?
Q：什麼是「Bank Less」？從數位溝通的角度來看，DBS 試圖向消費者傳遞什麼樣的品牌價值主張（Value Proposition）？
Q: To realize this promise, what changes did DBS make in product design?
Q：為了實現這個承諾，DBS 在產品設計上做了什麼改變？
Reveal the cost and risk that operating systems and infrastructure must bear in order to deliver an “extreme digital experience.”
揭示為了達成上述的「極致數位體驗」，企業在作業系統與基礎架構上必須付出什麼代價與風險。
Background: To achieve GANDALF-like agility and fast release capability, DBS broke a large monolithic system into a cloud-native microservices architecture and introduced hundreds of agile development pods. Yet between 2023 and 2024, DBS experienced at least six severe system outages.
背景：為了達到像大型科技公司（GANDALF）一樣的敏捷性與快速發布能力，DBS 將龐大的單體系統拆解為雲端原生的微服務架構（Microservices），並導入了數百個敏捷開發小組（Pods）。然而，在 2023 年至 2024 年間，DBS 發生了至少 6 次嚴重的系統中斷。
Q: To achieve agility and a seamless experience, DBS changed the backend operating system. Why did this transformation instead lead to a series of outages in 2023?
Q：為了做到敏捷與無縫體驗，DBS 改變了後台的作業系統。為什麼這種轉變反而導致了 2023 年的一連串當機？
Key takeaway: This reveals a coordination gap. As Piyush Gupta admitted: “I did not think enough about the operating complexity created by a distributed microservices architecture.” Although unbundling tasks lets individual teams release functions quickly, when the system becomes highly interdependent and complex, a single-node failure can easily trigger a chain reaction.
關鍵觀察：這裡可以看見「協調差距（Coordination gap）」。正如 Piyush Gupta 坦承：「我沒有充分思考分散式微服務架構所帶來的營運複雜性。」雖然解綁（Unbundling）任務能讓個別團隊快速推出新功能，但當系統變得高度關聯與複雜時，單一節點的故障很容易引發連鎖反應。
Q: Are these instabilities purely an IT problem, or a problem of business operating design?
Q：這些不穩定是純粹的「IT 技術問題」，還是「商業營運設計的問題」？
Key takeaway: This is not only an IT problem. It is the result of imbalance between the pursuit of innovation speed and the need for system resilience. Marketing and business units keep pushing for new features, while the carrying capacity of the infrastructure is ignored.
關鍵觀察：這不僅是 IT 問題。這是企業在追求「創新速度」與「系統韌性」之間失衡的結果。行銷與業務部門不斷推動新功能上線，卻忽略了基礎架構的乘載能力。
Bring the technical disaster back into the perspective of digital communication and marketing, then assess the damage to brand trust and the logic of crisis communication.
將技術災難拉回「數位溝通與行銷」的視角，評估系統不穩定對品牌信任的破壞，以及企業的危機溝通策略。
When outages occur, customers cannot pay or transfer money. The Monetary Authority of Singapore imposed heavy penalties, required an additional SGD 1.6 billion capital buffer, and suspended non-essential IT changes and new business acquisitions for six months. Piyush Gupta took a 30% pay cut, and senior executives took 21% pay cuts to show accountability.
當當機發生時，首當其衝的是無法結帳、無法轉帳的顧客。新加坡金管局（MAS）祭出重罰，要求增加 16 億新幣的資本緩衝，暫停非必要的 IT 變更與新業務收購 6 個月。Piyush Gupta 減薪 30%，高階主管減薪 21% 以示負責。
When “invisible banking” suddenly becomes visible and unusable, customer friction reaches its highest level. This destroys the core promise of being dependable. In the digital age, app uptime is the most important form of brand marketing.
當隱形銀行（Invisible banking）突然「現形」且無法運作時，顧客的摩擦力達到頂峰。這直接摧毀了「Dependable（可靠）」這個核心服務承諾。在數位時代，「App 的可用性（Uptime）就是最重要的品牌行銷」。
What irony and brand damage did the outages create for the line “Live More, Bank Less”?
系統當機對「Live More, Bank Less」這個品牌標語造成了什麼樣的反諷與品牌傷害？
From the perspective of a chief marketing and communications officer, how should DBS communicate digitally in order to restore brand equity during the 2023 trust crisis?
站在行銷與公關長的角度，面對 2023 年連續當機的品牌信任危機，DBS 應該如何進行數位溝通以挽回品牌權益？
The symbolic meaning of the CEO taking the lead in accepting a pay cut.
CEO 帶頭減薪的象徵意義。
Set aside SGD 80 million to improve system resilience and commit that if another outage occurs within six months, recovery time will not exceed three hours.
撥出 8,000 萬新幣專項預算提升系統韌性，承諾在 6 個月內若再當機，修復時間不超過 3 小時。
When digital channels fail, branches and withdrawal points must stay open—even on Sundays.
數位管道中斷時，保持分行與提款點暢通，甚至週日開門營業。
True omnichannel is not only for marketing. It is also an operating-resilience mechanism.
真正的全通路（Omnichannel）不僅是為了行銷，更是為了營運韌性。
Break the one-way myth that “digital transformation = technology adoption” and build a two-way view of evolution from both the supply side and the demand side across generations.
打破「數位轉型＝導入科技」的單向迷思，建立供給端與需求端（不同世代）的雙向數位化演進觀點。
Understand that excellent customer experience comes from end-to-end internal process reconfiguration, and that service resilience is directly equivalent to brand trust.
理解卓越的顧客體驗來自「端到端（End-to-End）」的內部流程重組，且服務韌性直接等同於品牌信任。
Use the concept of Reshuffle to anticipate the unbundling of workflows and the emerging mechanism of “Agent-to-Agent” coordination in the AI era.
導入《Reshuffle》觀念，預視 AI Agent 時代下，工作流程的解構與「Agent 對 Agent」的新協作機制。
We often mix up Digitization, Digitalization, and Digital Transformation. Here we map them not only from the enterprise supply side, but also from the consumer demand side and the traits of different generations. Digital evolution is jointly completed through interaction between firms and consumers.
我們常把 Digitization、Digitalization 與 Digital Transformation 混為一談。今天我們不只從企業（供給端）來看這三個字，還要從消費者（需求端）的行為，以及不同世代的發展特徵來做 Mapping。數位化演進，其實是企業與消費者在互動中共同完成的。
| Evolution level演進層次 | Enterprise supply-side view企業供給端觀點 | Consumer demand-side view消費者需求端觀點 | Generational mapping對應世代與互動特徵 |
|---|---|---|---|
| 1. Digitization1. Digitization（數位化／資訊轉化） | Data digitization: convert physical or analog data into digital formats, such as paper medical records into electronic files or physical forms into PDFs.資料的數位化：將實體紙本或類比資料轉換為數位格式（如：紙本病歷轉電子檔、實體表單轉 PDF）。 | Behavior goes online: actions that once required physical completion move online, such as viewing a product catalog on the internet or downloading coupons.行為的線上化：將原本必須實體完成的動作搬到線上（如：在網路上查看商品型錄、下載折價券）。 | Gen Y sees the internet as a tool. Online and offline behavior are clearly separated, and going online is for solving specific tasks.Gen Y（千禧世代）：將網路視為「工具（Tool）」。線上與線下行為有明顯界線，上網是為了解決特定任務。 |
| 2. Digitalization2. Digitalization（數位優化／流程數位化） | Process digitization: use digital technologies to improve existing business processes and raise efficiency, such as RPA automation or CRM systems.流程的數位化：利用數位技術改善現有業務流程，提升效率（如：導入 RPA 自動化流程、建立 CRM 系統）。 | Experience digitization: consumers expect smoother digital interaction, such as mobile payments, app-based appointments, or customer service inquiries through social media.體驗的數位化：期待更流暢的數位互動（如：使用行動支付、App 預約服務、透過社群媒體詢問客服）。 | Gen Z pursues strong personalization and interaction, shows a pragmatic attitude, values convenience, and builds a profile through digital footprints.Gen Z（Z 世代）：追求高度個人化與互動。展現務實態度，重視使用便利性，並透過數位足跡建立個人輪廓。 |
| 3. Digital Transformation3. Digital Transformation（數位轉型／商業重構） | Business model reconfiguration: develop new organizational capabilities and technologies to keep improving customer experience, lower unit cost, and create lasting competitive advantage.商業模式的重構：開發全新的組織能力與技術，持續改善顧客體驗並降低單位成本，創造持續性競爭優勢。 | Native digital / phygital fusion: behavior is no longer separated into online versus offline. Consumers expect seamless, immersive, and predictive value delivery.原生數位／虛實融合：行為不再區分線上線下。期待品牌提供無縫、沉浸式、預測性的價值交付。 | Gen Alpha are phygital natives. For them, the physical and digital worlds have no boundary. They are accustomed to AI-native interaction and adapt to virtual communities and metaverse environments.Gen Alpha（Phygital Natives）：「實體數位原住民」。對他們而言，實體與數位世界沒有邊界。習慣與 AI 原生互動（AI Natives），並適應虛擬社群與元宇宙環境。 |
Gen Y → Gen Z → Gen Alpha
Gen Y → Gen Z → Gen Alpha
Click to reveal what changes across generations.
點擊看見世代差異。
Tool use → convenience and personalization → seamless, AI-native, phygital experience.
工具使用 → 便利與個人化 → 無縫、AI 原生、虛實融合體驗。
Where does most customer service still stay today?
現在多數企業的客戶服務，停留在哪一個階段？
When a Gen Z or Gen Alpha consumer—expecting Digital Transformation or even AI-native interaction—meets a company that only achieves Digitization, such as putting a form online as a PDF, the result is a severe experience gap.
當一個 Z 世代或 Alpha 世代的消費者（期待 Digital Transformation 甚至 AI 原生互動）遇到一家只做到 Digitization（只把表單變成線上 PDF）的企業時，會產生明顯的體驗斷層。
Many people think customer experience is just whether the front-end app looks good. In reality, excellent experience depends on end-to-end internal digital process reconfiguration. If front-end innovation is not supported by the backend, fragility appears. DBS uses Managing through Journeys (MtJs) to break silos and connect cross-functional teams.
很多人以為「顧客體驗」只是前端 App 做得漂不漂亮，但實際上，卓越的體驗依賴內部「端到端（End-to-End）」的數位流程重組。前端的創新如果沒有後端基礎設施的支撐，就會產生脆弱性。星展銀行（DBS）就是透過「以顧客旅程為導向（Managing through Journeys, MtJs）」來打破部門穀倉，串聯跨功能團隊。
In the digital age, service availability and resilience are part of the core brand value itself. An outage is not only an IT incident but also a direct public-relations and trust crisis.
在數位時代，服務的可用性（Availability）與韌性本身就是品牌價值的核心。當機不僅是 IT 問題，更是直接的公關與信任危機。
Why can a small failure cause full paralysis? Because within the end-to-end process, marketing, pricing, customer service, and fulfillment are tightly connected.
為什麼一個小環節的當機會導致全面癱瘓？因為端到端流程中，從行銷、定價、客服到訂單履行是高度相連的。
Once a firm digitizes end-to-end processes, what comes next? The next step is to hand execution to AI agents. The true power of AI is not “automating a single task,” but changing the coordination mechanism of the system.
當企業把端到端流程數位化後，下一步是什麼？是交給 AI 代理（Agents）來執行。AI 的真正力量不在於「自動化單一任務」，而在於「改變系統的協調方式（Coordination）」。
Summary: From Digitization to Agentification, what changes is not only the technology stack, but the coordination mechanism. The priority is no longer just buying new software; it is reorganizing the firm so it can function in automated, cross-boundary coordination.
總結收斂：從 Digitization 到 Agentification，我們看到的不僅是技術升級，更是「協調機制（Coordination Mechanism）」的根本改變。轉型優先順序不再只是買一套新軟體，而是如何重構你的組織，讓它能夠適應這種高度自動化、跨邊界的新協作系統。
In this Tool Lab, you will use Deep Research tools, test their claims, and deliberately look for mistakes. The point is not to accept AI output at face value, but to practice verification, comparison, and critical judgment. The final segment uses a node representation so you can see the structure of an Agentic Workflow more clearly: trigger, intent recognition, tool calling, and human handoff.
在這個 Tool Lab 中，你要親自操作 Deep Research 工具、檢查它的主張，並刻意找出錯誤。重點不是照單全收 AI 的輸出，而是練習查證、比對與批判判斷。最後一個環節採用節點表示法（Node Representation），讓你更清楚看見 Agentic Workflow 的基本結構：觸發、意圖識別、工具呼叫與人類交接。
Master AI deep research tools such as Gemini Advanced and Perplexity Pro, and learn to identify hallucinations and logic gaps.
掌握 AI 深度研究工具（如 Gemini Advanced、Perplexity Pro），並具備辨識「AI 幻覺」與「邏輯縫隙」的能力。
Use the frameworks of Rewired and Reshuffle to diagnose the digital transformation of a real company.
利用《Rewired》與《Reshuffle》架構，對一家真實企業進行數位轉型診斷。
Prepare for Session 2 by drawing the first node-based customer-service / complaint-handling workflow.
為 Session 2 準備，繪製第一個客戶服務／客訴處理的節點式代理工作流程圖。
Deep Research tools may appear to produce a perfect industry report in one minute, but they have three fatal weaknesses: they confuse plans with achievements, misplace cause and effect, and mix up the timeline. Today’s task is to be the AI’s auditor.
現在市面上的 Deep Research 工具看似能在一分鐘內寫出完美的產業報告，但它們有三個致命傷：第一，錯把「計畫」當「實績」；第二，因果錯置；第三，時間線混亂。我們今天的任務，就是要當 AI 的「審查委員（Auditor）」。
Example: Nike raised the share of D2C revenue to 40% through the SNKRS App and substantially reduced dependence on distributors.
範例：Nike 透過 SNKRS App 成功將 D2C（直接面對消費者）營收比例提升至 40%，大幅降低對經銷商的依賴。
Click the AI’s footnotes and trace them back to the most original source, such as annual reports, press releases, or analyst reports.
點擊 AI 提供的引註（Footnotes），找出最原始的來源（如財報、新聞稿、分析師報告）。
Scoring rule: 1 = content farm / PR copy; 3 = business-news report; 5 = SEC filing / official statement. Example score: 4.
評分標準：1＝內容農場／公關稿；3＝商業新聞報導；5＝SEC 財報／官方聲明。範例評分：4。
This is the most important step. You must actively look for opposing evidence. Example: AI ignored the fact that in 2023–2024 Nike cut wholesale too aggressively, which led to excess inventory and revenue decline, and later returned products to Macy’s and Foot Locker. The AI claim is outdated or one-sided.
這是最重要的一步。你必須去尋找反面證據。範例：AI 忽略了 Nike 在 2023–2024 年因為過度切斷經銷商（Wholesale）導致庫存積壓與營收衰退，近期又重新將產品放回 Macy’s 與 Foot Locker 的事實。AI 的主張是過時或片面的。
Study Starbucks or Nike. Both firms are currently going through painful transformation restructuring. Starbucks faces store-flow breakdown caused by Mobile Order, along with a new CEO. Nike faces the cost of over-relying on D2C algorithms while becoming disconnected from physical distributors. The example here uses Starbucks.
研究 Starbucks（星巴克）或 Nike。這兩家企業近期都在經歷痛苦的轉型重組：星巴克面臨 Mobile Order 造成的門市動線崩潰與新 CEO 上任；Nike 面臨過度依賴 D2C 演算法與實體經銷商斷裂的代價。這裡以星巴克為例。
Use a master prompt that combines McKinsey’s four capabilities in Rewired—talent, operating model, technology, and data—with the “coordination gap” concept from Reshuffle. Feed the facts you have just verified into AI and generate a one-page diagnosis draft.
使用一個精心設計的 Master Prompt。這個 Prompt 融合了 McKinsey《Rewired》的四大能力（人才、營運模式、技術、數據）以及《Reshuffle》的「協調差距（Coordination Gap）」概念。請將剛才查證過的事實結合這個 Prompt 輸入給 AI，產出一頁式的診斷草稿。
[Transformation Diagnostic Prompt] You are a top-tier digital-transformation consultant. Based on the company information I provide (or the latest information you search), produce a one-slide digital-transformation diagnosis for Starbucks. Please output in bullet form along exactly these three dimensions: 1. Value Proposition - What was the company’s traditional value? - After digitalization (for example, Mobile Order & Pay and the membership app), how has the value proposition qualitatively changed? - What coordination gap has appeared? Example: frictionless online ordering causes the physical workflow of baristas in stores to break down, and customers wait much longer. 2. Key Processes - Identify one core process that has been most deeply affected by digitalization, such as ordering and fulfillment. - Which tasks in that process were automated? - What new pain points appeared? 3. Capability Gaps - Talent: what gap exists between front-line employees and the technology system? - Operating Model: is there a disconnect between the agile team at headquarters building the app and the teams running store operations? - Technology & Data: can the current system architecture support explosive real-time data volume? Output requirement: Convert the AI-generated text into a concise one-pager.【Transformation Diagnostic Prompt】 你現在是一位頂尖的數位轉型顧問。請根據我提供的企業資訊（或你搜尋到的最新資料），針對「星巴克（Starbucks）」進行一頁式的數位轉型診斷。 請嚴格按照以下三個維度進行條列式輸出： 1. Value Proposition（價值主張的重塑） - 該企業的傳統價值是什麼？ - 在數位化（如 Mobile Order & Pay、會員 App）介入後，價值主張產生了什麼質變？ - 產生了什麼「協調差距（Coordination Gap）」？（例如：線上無摩擦點餐，導致實體門市咖啡師的製作流程崩潰與顧客等待時間過長）。 2. Key Processes（關鍵營運流程的解構與重組） - 請指出一個被數位化影響最深的「核心流程」（如：點餐與訂單履行）。 - 該流程中的哪些任務被自動化了？ - 又產生了哪些新的痛點？ 3. Capability Gaps（四大底層能力缺口） - Talent（人才）：第一線員工與科技系統的適應落差？ - Operating Model（營運模式）：總部開發 App 的敏捷團隊與門市營運團隊是否有脫節？ - Technology & Data（技術與數據）：現有系統架構是否撐得起爆量的即時數據運算？ 產出要求： 將 AI 生成的文字，轉化為一頁精煉的投影片（One-Pager）。
In Session 2 we will discuss Salesbots and AI Agents. An agent is not a traditional one-question / one-answer bot. It is a workflow that can autonomously decide its route and use tools. We therefore practice a node-based workflow because mainstream frameworks such as LangGraph, n8n, and Vapi are graph- and state-machine-based. Each node represents an action or decision.
在下次的 Session 2，我們將探討 Salesbot 與 AI Agents。Agent 不是傳統的一問一答機器人，它是能夠「自主決定路徑與使用工具」的工作流。為了讓大家具象化，我們今天不寫程式，而是練習畫出節點式工作流（Node-based Workflow）。現在主流的 Agent 框架（如 LangGraph、n8n、Vapi）都是基於圖形（Graph）與狀態機（State Machine）設計的。每個節點代表一個動作或決策。
Customer sends a complaint email or types a message into the customer-service chat box.
客戶寄出抱怨 Email 或在客服對話框輸入訊息。
AI determines whether this is a general inquiry, a return request, or a highly angry complaint.
AI 判斷這是一般詢問、退貨請求，還是強烈憤怒的客訴。
Branch A: for returns, the agent queries the ERP system for order status. Branch B: for policy lookup, the agent calls a RAG system to search the company’s return policy.
分支 A（退貨）：AI 自動呼叫 ERP 系統查詢訂單狀態。分支 B（政策查詢）：AI 呼叫 RAG 搜尋公司退換貨政策。
The system checks whether the order amount is greater than USD 500, and whether the emotion score is at a critical level of anger.
系統判斷：此訂單金額是否大於 500 美金？情緒分數是否處於極度憤怒（Critical）？
If conditions allow automation, the agent automatically generates a return label and sends email. If a guardrail is triggered, the system packages the summary and routes it to a human support lead.
條件符合自動化：Agent 自動生成退貨標籤並寄送 Email。條件觸發護欄：進入 Human-in-the-loop，將所有彙整好的資料打包拋轉給真人客服主管。
Digitize this workflow (submit as PDF or screenshot) and mark the node at which AI is most likely to create risk—for example, promising the wrong compensation. This will connect directly to the trust and governance issue in “Should We Deploy a Gen AI Salesbot.”
將這張節點工作流程圖數位化（提交 PDF 或截圖），並標註在哪一個節點上，AI 最有可能產生風險（如承諾了錯誤的賠償）。這將銜接下次探討《Should We Deploy a Gen AI Salesbot》的信任與治理議題。
A seamless digital promise cannot stand on front-end rhetoric alone. If the backend architecture cannot carry the promise, the brand will absorb the damage first.
無縫的數位承諾不能只建立在前端說法上。若後端架構撐不起承諾，最先承受損害的就是品牌。
Customer experience is not just interface design. It is an end-to-end coordination problem that connects systems, teams, and physical operations.
顧客體驗不只是介面設計，而是連接系統、團隊與實體營運的端到端協調問題。
The next step after digitization is not merely more software. It is a redesign of coordination so that AI agents can operate responsibly across boundaries.
數位化之後的下一步，不只是更多軟體，而是重新設計協調機制，讓 AI Agent 能夠跨邊界且負責任地運作。