Working with AI comes in roughly three levels. First, we have people who are AI curious. This group relies on the free tier of AI tools and only uses chatbots

使用 AI 大致分為三個層級。首先，我們有對 AI 好奇的人。這群人依賴 AI 工具的免費版本，只有在別人提醒或遇到困難時才使用聊天機器人

when someone reminds them to or when they're stuck. Level two, we have the AI literate. These people pay for AI, maintain a prompts database, and they know when to use which AI feature

。第二級是 AI 素養者。這些人付費使用 AI，維護一個提示詞資料庫，而且他們知道什麼時候該用哪個 AI 功能

and model. Level three is AI native. And in a nutshell, these people have redesigned their workflows assuming an AI collaborator exists. Most professionals are stuck at level two.

和模型。第三級是 AI 原生者。簡單來說，這些人已經重新設計了他們的工作流程，假設有一個 AI 協作者存在。大多數專業人士卡在第二級。

So, in this video, I'll share the specific strategies ordered from simple to advanced that will get you to level three. Let's get started. Kicking things off with the easiest habit to adopt,

所以，在這支影片中，我會分享從簡單到進階排列的具體策略，讓你達到第三級。讓我們開始吧。首先是最容易養成的習慣，

leave AI breadcrumbs. What this means is instead of treating AI chats as disposable one-off threads that become almost impossible to find again, you want to create a hyperlink to the

留下 AI 麵包屑。這意味著，不要把 AI 對話當作用完就丟、幾乎不可能再找到的一次性對話串，你要創建一個超連結指向

conversation and paste it directly into the document where you're actually using the output. I know this habit sounds silly and insignificant, but it works great thanks to the core productivity

那個對話，然後直接貼到你實際使用輸出的文件中。我知道這個習慣聽起來很傻、微不足道，但它效果很好，因為符合核心生產力

principle of always organizing your information by where you will use it, not where you found it. Diving into a real world example, here's the actual Google Doc I used to prepare for a

原則：永遠按照你會用到它的地方來整理你的資訊，而不是按照你發現它的地方。來看一個實際的例子，這是我為一場

recent work presentation. The final outline tab has all my content and the helpful hints tab has hyperlinks to my AI conversations. Taking a step back, let's say I'm building this presentation

最近的工作簡報準備時用的實際 Google 文件。最終大綱分頁有我所有的內容，而有用提示分頁有指向我 AI 對話的超連結。退後一步來看，假設我從頭開始建立這個簡報

from scratch. I would first ask AI to rewrite my initial rough prompt so that it's optimized for the model that I'm using. And once I press enter, you'll

。我會先請 AI 重寫我最初粗略的提示詞，使它針對我使用的模型進行優化。一旦我按下 Enter，你會

notice the URL transforms into a unique link. Right? And this is where I'd press command or control L to select the entire URL, copy, come back to the

注意到 URL 變成了一個唯一的連結。對吧？這就是我會按 Command 或 Control L 選擇整個 URL、複製、回到

Google doc, type out Gemini, command or control K to hyperlink, and paste that link. I then copy the optimize prompt and paste that into a new chat. Make adjustments as needed.

Google 文件、打出 Gemini、按 Command 或 Control K 建立超連結、然後貼上那個連結的地方。然後我複製優化後的提示詞，貼到一個新的對話中。根據需要進行調整。

then go back and forth with the AI to brainstorm and refine my presentation outline. And of course, I save this new chat link in the Google doc as well, so

然後與 AI 來回討論，腦力激盪並完善我的簡報大綱。當然，我也把這個新的對話連結存到 Google 文件中，這樣

I can easily pick up where I left off a day or even a week later. Pro tip, add context next to each hyperlink so you remember why it matters. For example,

我可以輕鬆地從我一天甚至一週前停下的地方繼續。專業提示：在每個超連結旁邊加上上下文，這樣你就能記得它為什麼重要。例如，

here, this Gemini conversation was helping me brainstorm my outline. This one was around applying storytelling principles to that final outline. And I use Claw to refine my final talking

這裡，這個 Gemini 對話是在幫我腦力激盪大綱。這個是關於將說故事原則應用到最終大綱。然後我用 Claude 來完善我最終的

points. And just to be clear, I like to test multiple frontier models because I do this for a living. Most people should just pick one AI chatbot and get really

談話要點。要說清楚的是，我喜歡測試多個前沿模型，因為這是我的工作。大多數人應該只選一個 AI 聊天機器人然後真正

good at it. Here's another real world example. Whenever I create a new project page in Notion here under the op center section, I would add links to the corresponding CHP and claude projects so

熟練它。這是另一個實際的例子。每當我在 Notion 這裡的營運中心區塊下創建一個新的專案頁面，我會添加對應的 ChP 和 Claude 專案的連結，這樣

I can jump into those AI workspaces immediately. Put simply, leaving AI breadcrumbs means organizing your AI chats by work context and not by date or chronology. This beats trying to search

我可以立即跳進那些 AI 工作空間。簡單來說，留下 AI 麵包屑意味著按工作脈絡而不是按日期或時間順序來組織你的 AI 對話。這比試圖搜尋

for that one specific thread from days or weeks ago. And the rule of thumb here is simple. If the AI conversation took more than 10 minutes or produce something you'll reference again, anchor

幾天或幾週前的某個特定對話串要好得多。這裡的經驗法則很簡單：如果 AI 對話花了超過 10 分鐘，或產生了你會再次參考的東西，立刻把它

it to your workspace immediately. Speaking of systems, I'm actually building an entire course on evergreen AI skills that teaches universal principles for any platform. So, if you want a framework that never goes

錨定到你的工作空間。說到系統，我其實正在建構一個完整的課程，關於永不過時的 AI 技能，教授適用於任何平臺的通用原則。所以，如果你想要一個永不

obsolete, click the link below to join the weight list. Moving on to habit number two that requires a bit more effort, build an AI swipe file system.

過時的框架，點擊下方連結加入等待名單。接下來是需要更多努力的習慣二，建立一個 AI 靈感檔案系統。

In a nutshell, instead of prompting AI with basic instructions like write a business proposal, you provide a specific example from your curated library aka your swipe file and ask the

簡單來說，不要用基本指令提示 AI，像是「寫一份商業提案」，而是提供一個來自你策展資料庫的具體例子，也就是你的靈感檔案，然後請

AI to first analyze what makes it so effective. Then apply those patterns to your new content. For instance, let's say you work in OpenAI and you have this brilliant idea that you know users

AI 先分析是什麼讓它如此有效。然後將那些模式應用到你的新內容。例如，假設你在 OpenAI 工作，你有一個絕妙的想法，你知道全世界的用戶

around the world will absolutely love pumping chat full of ads. Instead of starting a proposal from scratch, you open up your AI swipe folder to find examples of business proposals you've

一定會超愛在 Chat 裡塞滿廣告。不要從頭開始寫提案，你打開你的 AI 靈感資料夾，找到你之前儲存的商業提案範例

previously saved. Share them with the AI and say, "Analyze the business proposals I've attached, list the key patterns in structure and tone, then apply those patterns to my content below." And you

。與 AI 分享它們並說：「分析我附上的商業提案，列出結構和語調的關鍵模式，然後將那些模式應用到我下面的內容。」然後你

paste your advertising product idea designed to maximize shareholder value.

貼上你為了最大化股東價值而設計的廣告產品想法。

Jokes aside, I guarantee you that initial output will be stronger than any initial draft you could have come up with yourself, not to mention the massive time savings. Funny story,

玩笑歸玩笑，我保證那個初始輸出會比你自己能想出的任何初稿都要好，更不用說節省大量時間。有趣的故事，

when I was at Google, I used this technique for all my important presentations by uploading slide decks from Mackenzie Bane and BCG. And senior leaders who came from those firms would

當我在 Google 的時候，我用這個技巧做所有重要的簡報，上傳麥肯錫、貝恩和 BCG 的投影片。那些來自這些公司的資深主管會

ask if I had also worked there before.

問我是否也曾在那裡工作過。

And I'd be like, "What? No, these frameworks and principles come so naturally to me. Does it not come naturally to you?" And that's why I'm not at Google anymore. No, just kidding.

我會說：「什麼？沒有，這些框架和原則對我來說太自然了。對你來說不自然嗎？」這就是為什麼我不在 Google 了。開玩笑的。

But as you can see, this technique is so effective because it gives the AI a clear picture of what good looks like, allowing Chachi or Gemini to produce output that matches those standards

但如你所見，這個技巧之所以有效，是因為它給了 AI 一個清晰的「好」是什麼樣子的圖像，讓 ChatGPT 或 Gemini 能產出符合那些標準的輸出

instead of generic slot. To close the loop, the actual habit you want to develop is whenever you encounter excellent work in your field, immediately save it to your swipe file

而不是通用的廢話。做個總結，你想培養的實際習慣是：每當你在你的領域遇到優秀的作品，立刻把它存到你的靈感檔案

system so you can reference it the next time you face a similar task. Pro tip: start narrow and expand gradually. Begin with just two to three use cases you do repeatedly like presentations, emails,

系統，這樣下次面對類似任務時你可以參考它。專業提示：從窄範圍開始，逐漸擴展。先從兩到三個你經常做的用例開始，像是簡報、電子郵件、

or reports. And organize your folders by use case, not by source or date. By the way, this is also the first step in making your Google Drive AI ready, which

或報告。按用例而不是按來源或日期來組織你的資料夾。順便說一下，這也是讓你的 Google Drive 為 AI 做好準備的第一步，這是

is something I dive deeper into in the course I just mentioned. And number three, we have AI first task planning.

我在剛才提到的課程中會更深入探討的內容。第三個習慣是 AI 優先的任務規劃。

Heads up, this habit is probably the hardest to maintain consistently, but I promise you, just like going to the gym, it will make a massive impact over the long run. Put simply, this

提醒一下，這個習慣可能是最難持續維持的，但我向你保證，就像去健身房一樣，從長遠來看它會產生巨大的影響。簡單來說，這個

habit involves planning your AI use before you start a big piece of work.

習慣涉及在開始一大塊工作之前規劃你的 AI 使用。

This means breaking down complex projects into small concrete tasks, then marking the ones AI can and should help with. Diving right into real world example, I used to be responsible for

這意味著把複雜的專案分解成小的具體任務，然後標記 AI 可以且應該幫忙的任務。直接進入實際例子，我曾經負責

sending these uh weekly newsletters to our Google Ads customers uh with the goal of driving adoption of new product features, aka getting them to spend more money. Basically, before writing

發送這些每週電子報給我們的 Google Ads 客戶，目標是推動新產品功能的採用，也就是讓他們花更多錢。基本上，在寫

anything, I'd break down the work into steps and microtasks. Then I decide whether to do each microtask manually or use AI. If I use AI, I specify the exact

任何東西之前，我會把工作分解成步驟和微任務。然後我決定每個微任務是手動做還是用 AI。如果用 AI，我會指定最

tool that's best suited for that task. In this case, there are three main steps. Step one is to clarify the goal and audience. Step two is to draft the

適合那個任務的確切工具。在這種情況下，有三個主要步驟。步驟一是釐清目標和受眾。步驟二是起草

newsletter. And step three is to refine the copy for Google's brand voice.

電子報。步驟三是為 Google 的品牌聲音完善文案。

Now, step 1.1 would be to brain dump key information like what the new feature is, the benefits, who should use it, and so on. This task is manual because there

現在，步驟 1.1 是腦力傾瀉關鍵資訊，像是新功能是什麼、好處、誰應該使用它等等。這個任務是手動的，因為有

are details I know that AI doesn't have access to, and I also want to inject my point of view. Step 1.2 is to fact check my notes from the previous step. Here,

一些我知道而 AI 沒有的細節，而且我也想注入我的觀點。步驟 1.2 是事實核查上一步的筆記。這裡，

it makes sense to use AI, specifically Notebook LM, since it has the lowest hallucination rates. And so I upload my brain dump and source documents onto a notebook to verify rollout dates,

使用 AI 是有意義的，特別是 Notebook LM，因為它的幻覺率最低。所以我把我的腦力傾瀉和來源文件上傳到筆記本，以驗證推出日期、

feature names, policy details, etc. Step 1.3 is to turn those fact check notes into a structured brief. This is obviously also a perfect task for AI, but this time I use the standalone

功能名稱、政策細節等。步驟 1.3 是把那些經過事實核查的筆記變成結構化的簡報。這顯然也是 AI 的完美任務，但這次我用獨立的

Gemini app instead of notebook LM because the Gemini app is much better at creative writing. In the interest of time, I'm going to skip over steps two and three, but the process is exactly

Gemini 應用程式而不是 Notebook LM，因為 Gemini 應用程式在創意寫作方面要好得多。為了節省時間，我跳過步驟二和三，但流程完全

the same. List out all the microtasks. decide if AI should help and if yes, pick the right tool for that specific job. At this point, the benefits should

一樣。列出所有微任務。決定 AI 是否應該幫忙，如果是，為那個特定工作選擇正確的工具。到這裡，好處應該

be pretty clear. First, you cut decision fatigue and context switching because AI usage is already preddecided at the task level. Second, you increase quality and speed by matching the right AI tool to

很明顯了。首先，你減少了決策疲勞和上下文切換，因為 AI 的使用已經在任務層級預先決定了。其次，通過將正確的 AI 工具匹配到

the right kind of work instead of forcing one tool to do everything. And the rule of thumb here is for any project that will take more than an

正確類型的工作而不是強迫一個工具做所有事情，你提高了品質和速度。這裡的經驗法則是，對於任何會花超過一個

hour, spend 5 to 10 minutes mapping the steps and tagging which ones are AI or manual. For the productivity nerds out there, this is a classic example of

小時的專案，花 5 到 10 分鐘規劃步驟並標記哪些是 AI 或手動。對於那些生產力狂熱者，這是一個經典的

sharpening the axe, where spending a few minutes on planning up front saves hours of work later. Pro tip, create templates for recording workflows, like what I have here for my weekly

磨斧子的例子，前期花幾分鐘規劃，後面省下好幾個小時的工作。專業提示：為記錄工作流程創建模板，就像我這裡為我的每週

newsletter, so that next time you can focus on executing instead of having to plan from scratch. This brings us to a bonus habit that ties everything together. Maintain a prompts database.

電子報做的一樣，這樣下次你可以專注於執行而不是從頭規劃。這帶我們到一個把一切串起來的額外習慣。維護一個提示詞資料庫。

I've talked about this many times before. Whenever you write a prompt that works well, save it to a central library organized by use case so you can reuse

我之前講過很多次了。每當你寫出一個效果很好的提示詞，把它存到按用例組織的中央資料庫，這樣每當你再次面對那個任務時你可以重複使用

that prompt whenever you face that task again. The worst feeling is writing a perfect prompt 3 weeks ago that generated a perfect output. But today you can't find it. So you try to rewrite

那個提示詞。最糟糕的感覺是 3 週前寫了一個完美的提示詞，產生了完美的輸出。但今天你找不到它了。所以你試著從記憶中重寫

it from memory and the result is just eh, which is also my girlfriend's reaction shortly after we started dating. Huh. A lot of you have been asking for my go-to prompts. So, I spent

它，結果只是還好，這也是我女朋友在我們開始約會後不久的反應。呵。很多人一直在問我常用的提示詞。所以，我花了

quite a bit of time putting together a set of essential prompts that everyone can benefit from because I genuinely believe you don't need a thousand random prompts. You need 10 to 15 battle tested

相當多的時間整理了一套每個人都能受益的必備提示詞，因為我真心相信你不需要一千個隨機的提示詞。你需要 10 到 15 個經過實戰檢驗的

ones that you can use every day. I've been using and refining these ever since Chachi PT first launched. So, if you want to skip the trial and error,

提示詞，你可以每天使用。自從 ChatGPT 首次推出以來，我一直在使用和完善這些。所以，如果你想跳過試錯過程，

I'll leave a link to this down below.

我會在下面留下連結。

See you on the next video. In the meantime, have a great one.

下支影片見。在此同時，祝你一切順利。