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
🇹🇼 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:
- 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。
- 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)
- 報價:從「老手喬半天」→ 填一筆、三十秒產出 PDF、版本不再混亂。
- 丈量:從「現場手量 + 回辦公室手畫 CAD」→ 手機掃描自動出 DXF、少一整段工。
- 銷售流程:從各憑本事的零散動作 → 一條有客戶資料庫、有狀態追蹤的閉環。
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:
- 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.
- 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)
- Quotes: from "a veteran negotiating for half a day" → fill one row, PDF in thirty seconds, no more version chaos.
- Measurement: from "measure on site + hand-draw CAD at the office" → phone scan, automatic DXF, an entire job removed.
- Sales process: from scattered, improvised actions → a closed loop with a customer database and status tracking.
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.