← back · Case 01 of 4

把 AI 落地到一個「沒有工程師」的傳統產業
Landing AI in a Traditional, Non-Digital Industry

🔴 去識別化:主詞用「某傳統建材通路商」、不露公司名 / 客戶 / 內部資料。數字只用真實可驗證的流程改善。

🇹🇼 中文

一句話

在一家幾乎全靠紙本與人工的傳統建材通路、我把「一份要喬半天的報價單」變成「填一筆、三十秒產出」、還把「現場手量、回辦公室手畫 CAD」變成「拿手機掃一掃,圖自動出來」。

問題

傳統產業最難的、不是技術、是那裡沒有工程師、流程全長在人身上

報價靠老手憑經驗算、靠 Excel 一格一格填;丈量靠捲尺量完回辦公室再手畫 CAD;客戶資料散在每個業務各自的本子裡。

每一步都「能動」、但每一步都吃人、都慢、都會錯。導入數位化的最大障礙、是這裡的人沒時間、也沒人懂工具。

怎麼用 AI 落地

我沒有要他們「全面數位轉型」——那會失敗。我挑了最痛、最高頻的兩個點、各做一條閉環:

  1. 報價自動化:把報價邏輯做成一套模板 + 腳本。業務在一張表填客戶與品項、三十秒內自動產出 PDF 報價單、自動落地、自動編版本號、手機端也看得到狀態。把整條銷售鏈(調研 → 洽談 → 出報價 → 回簽 → 通知施工)串成一條、客戶資料第一次有了統一的資料庫。
  2. 丈量自動轉 CAD:用 iPhone 的 LiDAR 掃描房間、自動輸出 DXF 檔、AutoCAD 直接打開就能出圖。原本「現場量 + 回辦公室手畫」的兩段工、壓成一次掃描。

關鍵不是工具多炫、是我懂這個產業的流程——知道哪一步是真痛點、哪一步動了會被現場排斥。這是外來的工程師做不到的。

量化結果(真實、可驗證)

這個案例證明什麼

我能把 AI 落地到最不數位、最沒有工程師、流程全靠人的真實營運場景——而且落地的點、是我用十幾年產業經驗挑出來的。AI 落地到傳統產業、難的從來不是技術、是有沒有人同時懂這行的流程、又懂怎麼讓機器接手。我剛好兩邊都站。

🇬🇧 English

One-liner

At a traditional building-materials distributor that ran almost entirely on paper and manual work, I turned "a quote that took half a day to negotiate" into "fill one row, get it in thirty seconds," and turned "measure on site, then hand-draw the CAD back at the office" into "scan it with a phone, and the drawing comes out by itself."

The Problem

The hard part of a traditional industry isn't technology — it's that there are no engineers, and every process lives inside a person.

Quotes are calculated by veterans from experience, filled cell by cell in Excel; measurements are taken with a tape measure and then hand-drawn into CAD back at the office; customer records sit scattered in each salesperson's own notebook.

Every step "works," but every step is human-bound, slow, and error-prone. The biggest barrier to digitizing isn't the tech — it's that the people here have no time, and no one who understands the tools.

How I Landed AI on It

I didn't push a "full digital transformation" — that fails. I picked the two most painful, highest-frequency points and built a closed loop for each:

  1. Quote automation — turned the quoting logic into a template plus a script. A salesperson fills in customer and items on one sheet, and a PDF quote is generated automatically within thirty seconds, auto-saved, auto-versioned, with status visible on mobile. The whole sales chain (research → negotiation → quote → signed return → notify construction) became one connected flow, and customer data had a single database for the first time.
  2. Measurement-to-CAD — scan a room with the iPhone's LiDAR and export a DXF automatically, ready to open straight in AutoCAD. What used to be two jobs ("measure on site" + "hand-draw at the office") collapsed into a single scan.

The point wasn't a flashy tool — it was that I understand this industry's workflow: which step is the real pain, and which step, if touched, gets rejected by the people on the floor. An outside engineer can't do that.

Measurable Results (real, verifiable)

What This Case Proves

I can land AI in the most un-digital, engineer-free, fully human-run operation there is — and the points where I landed it were chosen with a decade-plus of industry experience. Landing AI in a traditional industry was never about the technology; it's about whether someone understands both the trade's workflow and how to hand it to a machine. I happen to stand on both sides.

Next: Case 02 — A Non-Engineer Who Shipped 30+ Apps →