The day Claude Code changed how I work

I remember the day I started using Claude Code. It was a Tuesday. My team had one week to ship a website. I was the only developer. I felt worried. Before Claude Code, I wrote all my code myself. Small tasks took hours. Big tasks took days. I worked late. I was tired all the time. That Tuesday, I tried Claude Code. I asked it to help me build a login page. It wrote the code. I read it. I changed two small things. I was done in ten minutes. The next day, I asked it to make the database. It did. I checked the code. It was clean and clear. I learned something new from it too. By Friday, I finished the whole website. I was not tired. I had time to play with my kids. That week taught me something big. The best AI tools do not take your job. They make your day easier. You still make the choices. You still know the product. But you move faster. You worry less. Now I use Claude Code every day. I still write code. But I think more about what to build, not just how. My work feels bigger, not smaller. I think many people will feel like this soon. Not scared of AI. Thankful for it. It is a good day to be a builder.

我記得第一次用 Claude Code 的那一天。那是一個星期二。我們團隊有一個星期要做完一個網站。我是唯一的工程師。我很煩惱。 在那之前，我的程式都自己寫。小事花好幾個小時。大事要好幾天。我很常加班。一直都很累。 那個星期二，我試試看 Claude Code。我請它幫我做一個登入頁。它寫了程式。我看過、改了兩個小地方。十分鐘就做完了。 第二天，我請它做資料庫。它也做了。我看過，程式很乾淨、也很清楚。我還從裡面學到一些新的寫法。 到星期五，整個網站就做完了。我沒有很累。還有時間陪小孩。 那一週教我一件重要的事。好的 AI 工具不會搶你的工作。它會讓你的日子輕鬆一點。你還是做決定。你還是最懂你的產品。但你走得更快，也比較不用擔心。 現在我每天都用 Claude Code。我還是會寫程式。但我想的比較多是要蓋什麼，不只是怎麼蓋。我的工作感覺變大，不是變小。 我想很多人很快也會這樣。不怕 AI。反而感謝它。這是一個很好當工程師的時代。

The day Claude Code changed how I work was not dramatic. There were no fireworks, no press release. It was a quiet Tuesday afternoon in my small office, and I was staring at a deadline I didn't know how to meet. The project needed a backend, a frontend, and a database — the work of three developers — and I was one person with five days. A colleague had mentioned this tool, Claude Code, the week before. I had been skeptical. I had tried many AI coding tools before, and most of them produced things that looked right but broke in small, annoying ways. Still, with nothing to lose, I opened the terminal and typed a simple request: design the database for a simple event booking system. Within a minute, Claude had not only written the schema, but explained every decision — which fields it normalized, why it used one naming convention over another, and which edge cases it had anticipated. I read it slowly, like I was reading a good colleague's pull request. It was better than what I would have written alone, and I learned something from reading it. Over the next three days, I shipped that entire project. But the real change wasn't speed. It was that I started thinking differently about my own role. I was no longer the person who typed code all day. I was the person who decided what was worth building, tested the edges, asked the hard questions, and made the product feel right. The keystrokes were a tool, not an identity. Since that week, my calendar looks different. More time for thinking. More time for my family. Less time feeling stuck. That Tuesday, quietly, something important changed.

Claude Code 改變我工作方式的那一天，其實沒什麼戲劇化的畫面。沒有煙火，也沒有記者會。那是一個安靜的星期二下午，我坐在我那小小的辦公室裡，盯著一個我根本不知道怎麼趕得上的 deadline。那個專案要前端、要後端、要資料庫——三個工程師的工作量——而我只有一個人，五天。 前一個禮拜，有個同事跟我提過 Claude Code。我一開始是懷疑的。之前我試過很多 AI 寫程式工具，大部分產出的東西看起來對，但跑起來會在一些莫名其妙的地方壞掉。反正也沒什麼可以失去的，我打開終端機，試著打了一個很簡單的問題：幫我設計一個活動報名系統的資料庫。 一分鐘之內，Claude 不只寫出了 schema，還把每一個決策說給我聽——哪些欄位它做了正規化、為什麼選這個命名、哪些邊界情況它已經先想到了。我慢慢讀，像在看一個好同事的 pull request。那個版本比我自己寫的還好，而且我從讀它的過程裡學到東西。 接下來三天，我把整個專案交付了。可是真正的改變不是速度，是我開始用不一樣的眼光看自己在做什麼。我不再是整天在敲程式碼的那個人，我變成那個決定什麼值得做、測試邊界、問困難問題、讓產品長得對的人。敲鍵盤只是一個工具，不是我的身份。 從那週之後，我的行事曆變了。更多時間思考。更多時間陪家人。更少時間卡關。那個星期二，有一件重要的事情，悄悄地變了。

The day Claude Code changed how I work wasn't marked by any particular revelation — no cinematic moment of AI ushering in a new age. It was a grey Tuesday afternoon, and I was staring at a Gantt chart that had ceased to correspond to reality. A three-developer project had arrived on my desk with one developer — me — and a five-day window. In another era, I would have started negotiating scope. Instead, almost absent-mindedly, I opened a terminal and began a conversation that would quietly rearrange my professional identity. What surprised me was not Claude's speed. Speed had been the promise of every AI coding tool I'd tried in the preceding years, and speed, taken alone, tends to produce velocity without direction. What surprised me was judgment. When I asked for a database schema, Claude didn't just emit one — it explained which normalization choices it had made, where ambiguity lived in my brief, and which future changes it had anticipated. It was, for the first time, talking to me as a colleague rather than performing as a tool. Over that week I shipped three services, two integrations, and a deployment pipeline I would previously have considered over-ambitious. But the lasting shift wasn't productivity. It was the sense that my role had quietly moved up one abstraction layer. I wasn't the person who typed code anymore. I was the person who made architectural decisions, interrogated assumptions, shaped the user experience, and took responsibility for whether the thing existed at all. Typing, it turned out, had been the least of my contributions all along — I just hadn't noticed. This, I think, is the real story of the moment we're living through. The tools are not replacing us. They are revealing how much of what we called "work" was actually just friction — and how much of what's left is distinctly, irreducibly human. On that quiet Tuesday, I learned something about myself that I hadn't needed to know before: what I am actually for.

Claude Code 改變我工作方式的那一天，沒有什麼特別的啟示——不是那種電影式的「AI 新時代降臨」場景。那是一個灰灰的星期二下午，我看著一張甘特圖，它早就已經跟現實失聯。一個原本需要三位工程師的專案，落到我這個一個人的桌上，窗口只有五天。換成以前，我大概已經開始跟客戶談縮減範圍。那次我幾乎是心不在焉地打開終端機，開始了一場事後回頭看、悄悄重整了我職業身份的對話。 讓我意外的不是 Claude 的速度。速度，是我在那幾年試過的每一個 AI 寫程式工具都在承諾的東西，而光有速度，往往產生的是有方向感很差的高效率。真正讓我意外的是判斷力。當我要一個資料庫 schema，Claude 不只丟一份給我，還解釋它做了哪些正規化、我的需求裡有哪裡模糊、哪些未來可能的變動它已經先考慮進去了。那是我第一次感覺 AI 工具是在以同事的身份跟我對話，不是以工具的姿態替我做事。 那一個星期，我交出了三個服務、兩個整合、一條我以前大概不敢接的部署 pipeline。但真正留下來的，不是產出，而是我的角色悄悄往上搬了一層抽象。我不再是那個整天敲鍵盤的人，而是那個做架構決定、質疑假設、形塑使用者體驗、對這東西到底要不要存在負責的人。敲鍵盤，原來一直都是我的貢獻裡最小的那一塊——只是我以前沒察覺。 我覺得這才是我們現在正在經歷的時代的真正故事。工具不是在取代我們，是在把我們真正是為了什麼而存在顯現出來。那個安靜的星期二，我學到一件我以前不需要知道的事：我，到底，是做什麼用的。