案例集｜New 3 Limited

01

U Design Week 2023U Design Week 2023

情境：設計事業部獨立運營後，需要一個對外主舞台，把設計供應鏈、空間與生態能力說清楚，而不是再堆一場活動秀。

判斷：把 U Design Week 當事業部敘事與生態接口——先講清「誰來、為什麼來、帶什麼走」，活動是載體不是目的。

結果：形成可對外同步的方案主線（主舞台＋生態）；後續空間與供應鏈敘事可沿同一判斷延展。

Context: After the design business unit began operating independently, it needed an outward stage to make the design supply chain, spatial work and ecosystem legible—not another event for its own sake.

Judgment: Treat U Design Week as the unit’s narrative and ecosystem interface—first clarify who comes, why, and what they leave with; the programme is a carrier, not the goal.

Result: A syncable proposal spine (main stage + ecosystem); later spatial and supply-chain stories can extend from the same judgment.

02

雙十一品牌策略Double 11 brand strategy

情境：大促品牌策略要向集團同步——多業務線、多觸點，容易變成素材庫，而不是一條可執行的判斷。

判斷：先鎖「今年這場大促要讓人記住什麼、什麼不做」，再分配創意與規範。同階段的集五福、平昌／阿里奧運 VIS 等節點品牌，並入副句訓練，不單開卡。

結果：形成可向集團同步的品牌策略口徑；創意執行有邊界，節點品牌經驗可複用到後續發佈。

Context: Festival brand strategy had to sync across the group—many lines and touchpoints, easy to become an asset dump instead of one executable judgment.

Judgment: First lock what this year’s festival must be remembered for, and what not to do, then allocate creative and standards. Same-stage node brands such as Alipay’s 集五福 and PyeongChang / Alibaba Olympic VIS sit in the subtitle as training, not separate cards.

Result: A group-syncable brand-strategy brief; creative work had edges, and node-brand judgment carried into later launches.

03

某建築招標企業A construction bidding company

情境：開標分析靠刷站與 Excel，熟手一個項目要半天到一天。

判斷：先沉澱老師傅計算鏈成技能，不要先做漂亮的分析頁。

結果：約 1 分鐘完成；丟文件即用。

Context: Bid-opening analysis relied on scraping sites and Excel; a skilled person needed half a day to a day per project.

Judgment: First turn the veteran calculation chain into a reusable skill; do not start with a prettier analysis page.

Result: About one minute per project; drop in files and use.

04

某醫學 AI 產品A medical AI product

情境：已有對話入口與循證敘事，工作流仍像舊工具的功能樹。

判斷：不要做成「用了 AI 的醫學工具」，要以臨床決策為目標做 AI 原生。

結果：給出可上線的判斷標準與改版優先序。

Context: Chat entry and evidence narrative existed; the workflow still followed an old feature tree.

Judgment: Do not ship “medicine with AI bolted on”; design AI-native around clinical decisions.

Result: Ship criteria and a revision priority order.

05

某 AI×鏈上企業An AI × on-chain company

情境：香港 AI 研發企業要把鏈上能力，從財務副業接到研發與商業協作。

判斷：區塊鏈應升級為協作基礎設施，治理先行、可控試點，而不是擴散注意力。

結果：明確 P0（Agent 支付沙盒、存證、儀表盤）與「不做」清單。

Context: A Hong Kong AI R&D company needed on-chain capability to move from finance side-work into R&D and commercial collaboration.

Judgment: Upgrade blockchain to collaboration infrastructure; govern first, pilot under controls, do not scatter attention.

Result: Clear P0 (Agent payment sandbox, attestation, dashboard) and a Don't list.

06

某保險經紀產品上架An insurance brokerage product listing

情境：上新規則散落各處，業務與技術多輪翻譯，老系統不敢動。

判斷：讓 AI 讀懂舊庫，業務用熟悉規格直接發布，不推倒重建。

結果：上架週期從約 20 天壓至約 2 天。

Context: New-listing rules were scattered; business and engineering needed many translation rounds; the legacy system felt untouchable.

Judgment: Let AI read the old store so business can publish in familiar specs; do not rip and replace.

Result: Listing cycle fell from about 20 days to about 2 days.

07

某連鎖餐飲業務自助A chain restaurant, business self-service

情境：約八成標準需求仍走完整研發，排期極長。

判斷：業務用自然語言生成標準頁，技術只兜底兩成核心。

結果：80% 以上需求可自助完成。

Context: About 80% of standard requests still went through full engineering, with a very long backlog.

Judgment: Let business generate standard pages in natural language; engineering only covers the roughly 20% core.

Result: Over 80% of requests completed through self-service.

案例集Cases