Unit 4 | 13:00-13:45
Case Discussion: Customer Signal Agent Platform
案例討論：Customer Signal Agent Platform
The case asks one practical question: how can a company turn scattered customer signals into decisions that product, sales, support, and management can act on?
這個案例問的是一個實務問題：公司如何把分散的客戶訊號，整理成產品、業務、客服與主管都能採取行動的判斷？
Case Setup
The company already has signals. It lacks a working memory.
公司已經有訊號，缺的是可運作的共同記憶。
Sales notes, support tickets, meeting transcripts, product feedback, quotes, and competitor comments exist in different places. The platform organizes what the company already sees into business insight.
業務紀錄、客服單、會議逐字稿、產品回饋、報價資料與競品消息散在不同地方。平台的價值在於整理公司早就看見、卻沒有串起來的訊號，形成可討論的商業洞察。
Case Questions
Five questions keep the discussion concrete.
五個問題，讓討論保持具體。
Use the case to extract a design method students can reuse in their own companies.
請從案例中抽出一套設計方法，讓學員能帶回自己的公司使用。
- Which management problem does the platform solve?
- Which data can be read automatically, and which data needs confirmation?
- Who is harmed if the agent makes a wrong recommendation?
- Where should the first MVP stop?
- How will the team measure whether the platform is useful?
- 這個平台真正解決哪一個管理問題？
- 哪些資料可以自動讀取，哪些資料需要確認？
- 如果 Agent 建議錯誤，誰會受到影響？
- 第一個 MVP 應該做到哪裡就停？
- 團隊如何衡量平台是否有用？
Side Case Lenses
副案例視角
Starbucks/NomadGo reminds us that field adoption matters. Ramp shows how workers become builders. Palantir AIP shows why enterprise agents need permissions, audit, and high-value use cases.
Starbucks/NomadGo 提醒我們，現場採用比展示效果更重要。Ramp 說明知識工作者如何成為 builder。Palantir AIP 則提醒企業級 Agent 需要權限、稽核與高價值場景。