T
載入中...
載入中...
Diving right in, we have the prompt reversal technique for Chacha BT. And here's what it does. You know that back and forth we all have with AI. After
直接進入主題，我們有 ChatGPT 的提示反轉技巧。它的作用是這樣的。你知道我們和 AI 的那種來回，在
that initial prompt, we get a first result that's maybe 50% of what we want.
初始提示後，我們得到的第一個結果大約是我們想要的 50%。
So, we ask for some changes. The second result gets us to 60% and this goes on and on until we finally land on something that's about 90% of what we
所以，我們要求一些修改。第二個結果讓我們達到 60%，然後這樣持續下去，直到我們最終得到大約 90% 我們
wanted, right? What if we could skip all that back and forth and just jump straight to the near-perfect result every single time? That's what prompt reversal does. Think of it like cooking
想要的結果，對吧？如果我們可以跳過所有那些來回，每次都直接跳到接近完美的結果呢？這就是提示反轉的作用。把它想像成沒有食譜的烹飪
without a recipe. You add some salt, taste it, add some pepper, and keep tweaking until the dish is perfect.
。你加一些鹽，嚐一嚐，加一些鬍椒，不斷調整直到菜餚完美。
Right? Prompt reversal is like having a robot scan that final dish and producing the exact recipe we needed from the start so that next time we nail it in
對吧？提示反轉就像讓一個機器人掃描那道最終的菜，從一開始就產生我們需要的確切食譜，這樣下次我們就能一次就做對
one shot. Let's go through a real world example so you see exactly how this works. I start off by asking Chachi BT to analyze our main competitor, Anthropic, and walk me through their
。讓我們透過一個真實世界的例子，讓你看看這到底是怎麼運作的。我首先請 ChatGPT 分析我們的主要競爭對手 Anthropic，並帶我瞭解他們的
business strategy. I receive a comprehensive overview of Anthropic's entire business. But this is way too dense. I'm not sure what I should focus on. So, I course correct and I say,
商業策略。我收到了 Anthropic 整個業務的全面概述。但這太密集了。我不確定應該專注於什麼。所以，我修正方向說，
"Hey, this is too dense. Restructure this into the SWAT analysis format.
「嘿，這太密集了。把這個重新組織成 SWOT 分析格式。」
Strengths, weaknesses, opportunities, and threats. I want three bullet points per section and use clear and simple words." and I get a targeted SWAT analysis. But now this is too concise.
優勢、劣勢、機會和威脅。我想要每個部分三個要點，使用清晰簡單的詞語。」然後我得到了一個有針對性的 SWOT 分析。但現在這太簡潔了。
The bullet points need more detail. So I refine again. Hey, uh, this is now too concise. Flesh out each bullet point and add a subheading under each section called our strategic response and
要點需要更多細節。所以我再次細化。嘿，呃，這現在太簡潔了。把每個要點展開，並在每個部分下新增一個叫「我們的策略回應」的子標題，並
suggest one concrete action that we can take. Finally, I get exactly what I want. Strengths, three supporting bullet points, recommended action. Weaknesses, three subordinating bullet points, recommended action. Same thing for
建議一個我們可以採取的具體行動。最後，我得到了我想要的。優勢，三個支援的要點，推薦行動。劣勢，三個次要的要點，推薦行動。機會和
opportunities and same thing for threats. Perfect. Now, here's where the magic happens. Instead of accepting this result and moving on, I add one final prompt. Reverse engineer our conversation and write the single prompt
威脅也是同樣的結構。完美。現在，這裡是魔法發生的地方。不是接受這個結果然後繼續，我新增一個最終提示。反向工程我們的對話，寫出那個
that would have produced my final response in one go. And look at this.
能一次產生我最終回應的單一提示。看看這個。
Chatbt outputs the reverse engineered prompt in a code block for easy copy pasting. And we can test this out. I'm just going to copy this code, open up a
ChatGPT 在程式碼塊中輸出反向工程的提示，方便複製貼上。我們可以測試一下。我只要複製這段程式碼，開啟一個
new chat, paste it in, click enter, and I'm just going to fast forward this part here.
新的對話，貼進去，按 Enter，我會快進這部分。
And as you can see, we got that perfect detailed analysis in one step. That's pretty awesome. Not only does this technique save us so much time and
如你所見，我們一步就得到了那個完美詳細的分析。這真的很棒。這個技巧不僅長期為我們節省了大量時間和
effort over the long run, but it also captures all those little details we refined along the way. And it helps us become better prompt writers because we can actually see how optimized prompts
精力，它還捕捉了我們一路上細化的所有小細節。而且它幫助我們成為更好的提示編寫者，因為我們實際上可以看到最佳化的提示是如何
are structured. Pro tip, people often ask me which prompts I save to my prompts database in Notion. And honestly, most of these prompts come from the prompt reversal technique
結構的。專業提示，人們常問我把哪些提示儲存到我在 Notion 的提示資料庫。說實話，這些提示大多來自提示反轉技巧，
because those are the best prompts I use repeatedly. Speaking of writing better prompts, today's sponsor, HubSpot, has an awesome resource I'm pretty impressed with. It's their free ebook for professionals, Supercharge Your Workday
因為那些是我反覆使用的最好的提示。說到寫更好的提示，今天的贊助商 HubSpot 有一個很棒的資源，我印象深刻。這是他們為專業人士準備的免費電子書，《用 ChatGPT 提升你的工作日》
with ChachiBT. Longtime viewers know I made an entire video on Google Gemini's guide for professionals. And while that was a good start, the HubSpot ebook goes way deeper into actionable strategies
。長期觀眾知道我做過一整支關於 Google Gemini 專業人士指南的影片。雖然那是一個好的開始，HubSpot 的電子書更深入地探討了
for specific corporate roles. In fact, their section on prompt databases inspired a popular tip for my corporate workshops, which is to assign one single gatekeeper per team to add, refine, and
針對特定企業角色的可行策略。事實上，他們關於提示資料庫的部分啟發了我企業培訓中的一個熱門技巧，就是為每個團隊指定一個把關者來新增、細化和
remove templates from a shared prompts database. I'll leave a link to the free ebook in the description, and thank you HubSpot for sponsoring this portion of the video. All right, moving on to hack
從共享提示資料庫中移除模板。我會在描述中留下免費電子書的連結，感謝 HubSpot 贊助這部分影片。好的，進入第
number two, a technique I call the fiveinut amplifier. diving right into a real world example. When I was a marketing manager at Google, I had to rely on the product and sales teams to
二個技巧，我稱之為五分鐘放大器。直接進入一個真實世界的例子。當我在 Google 擔任行銷經理時，我必須依賴產品和銷售團隊來
provide content for my campaigns. I'd be like, "Hey, you're going to send me those slides tomorrow, right?" And they'd be like, "Oh, yeah, for sure.
為我的活動提供內容。我會說，「嘿，你明天會把那些幻燈片發給我，對吧？」他們會說，「哦，是的，當然。」
Promise." A week later, I'd be like, "Yo, where are the slides?" And they'd be like, "Oh, I'm so sorry. I was busy pitching clients and generating shareholder value, something you
「保證。」一週後，我會說，「嘿，幻燈片在哪裡？」他們會說，「哦，我很抱歉。我忙著向客戶推銷和創造股東價值，這是你們
marketing folks won't understand." And I'd be like, "Oh, go yourself." Just kidding. I love the sales team.
行銷人員不會理解的事情。」我會說，「哦，去你的。」開玩笑的。我愛銷售團隊。
Anyways, with ChadBt, as long as I can get my hands on their main slide deck, I can amplify that content all by myself.
總之，有了 ChatGPT，只要我能拿到他們的主要幻燈片，我就可以自己放大那些內容。
For example, I can ask the AI to create an engaging 10 question quiz based on their slides with multiple choice options and the correct answer indicated. Boom. Now, we can test the
例如，我可以要求 AI 根據他們的幻燈片建立一個有吸引力的 10 題測驗，有選擇題選項並標明正確答案。嘣。現在，我們可以測試
audience's knowledge and keep them engaged. Next, I'd have Catchup draft an internal recap email for stakeholders who couldn't attend the event, summarizing the key takeaways and product updates. Then, I might ask the
觀眾的知識並保持他們的參與。接下來，我會讓 ChatGPT 為無法參加活動的利益相關者起草一封內部回顧郵件，總結關鍵要點和產品更新。然後，我可能會要求
AI to create an external client-f facing infographic that pulls out the most impactful stats from those original slides. And now, I have a follow-up asset to share with clients. And this
AI 建立一個對外的客戶資訊圖表，從那些原始幻燈片中提取最有影響力的統計資料。現在，我有了一個後續資產可以與客戶分享。而且這
works across any department or role, right? The sales team can turn an industry report from marketing into cold emails for client outreach, a LinkedIn post to generate leads and a list of
適用於任何部門或角色，對吧？銷售團隊可以把行銷的行業報告轉換成客戶開發的冷郵件、產生潛在客戶的 LinkedIn 貼文和
talking points for their next call. HR can turn a 1-hour webinar transcript into a quick reference guide with step-by-step instructions, an FAQ for the company internet, and a knowledge
下次通話的談話要點。人力資源可以把一小時的網路研討會文字稿轉換成帶有步驟說明的快速參考指南、公司內網的常見問題，以及
check quiz for attendees. The benefits should be pretty obvious by now. We basically save hours of manual reformatting and rewriting by using AI to repurpose one higheffort piece of
參加者的知識測驗。好處現在應該很明顯了。我們基本上透過使用 AI 重新利用我們已有的一個高投入內容，節省了數小時的手動重新格式化和重寫
content we already have access to. Pro tip, be selective about your source material and only amplify what I call pillar content, your highquality proven work like a successful presentation or a
。專業提示，對你的源材料要有選擇性，只放大我所說的支柱內容，你高品質的經過驗證的工作，比如成功的簡報或
datari report because the AI will magnify the quality of what you give it.
資料包告，因為 AI 會放大你給它的品質。
Garbage in, garbage out, right? If you find my AI videos helpful, by the way, I'm actually developing an entire course on evergreen skills and universal principles to master any AI platform,
垃圾進、垃圾出，對吧？順便說一下，如果你覺得我的 AI 影片有幫助，我實際上正在開發一整門關於常青技能和通用原則的課程，幫助你掌握任何 AI 平臺，
giving you a futurep proof framework that never becomes obsolete. If you're interested in learning a practical and timeless AI system, click the link below to join the weight list. All right,
給你一個永不過時的未來防護框架。如果你有興趣學習一個實用且永恆的 AI 系統，點選下面的連結加入等待名單。好的，
moving on to strategy number three, the red team technique. This is a simple two-step method. First, you ask the AI to create something from your perspective. Then, and here's the
進入策略三，紅隊技巧。這是一個簡單的兩步方法。首先，你讓 AI 從你的角度建立一些東西。然後，這裡是
important part, you immediately ask it to flip the script and adopt a critical and opposing persona. Let's go through some examples. First up, job applications. You first ask Chashib to
重要的部分，你立即讓它翻轉劇本並採用一個批判性的對立角色。讓我們看一些例子。首先是求職申請。你先讓 ChatGPT
tailor your resume for a job description. Instead of stopping there, you then flip the script by following up with, "Now act as a hiring manager for this role. You're extremely busy and
為一個職位描述客製化你的履歷。不是停在那裡，你然後透過跟進來翻轉劇本，「現在扮演這個角色的招聘經理。你非常忙，只有
only have 60 seconds to scan the resume you just helped me write. What are your immediate red flags?" Example two, after asking AI to draft a business proposal for your CFO, you
60 秒來掃描你剛幫我寫的履歷。你立即的紅旗是什麼？」例子二，在讓 AI 為你的財務長起草一份商業提案後，你
redte team that proposal with you are now the company's chief financial officer CFO and your primary goal is to cut unnecessary costs. Read the proposal you just generated and critique it.
用「你現在是公司的財務長，你的主要目標是削減不必要的成本。閱讀你剛生成的提案並批評它」來對那個提案進行紅隊測試。
What's the biggest financial risk? Why isn't the ROI justified?
最大的財務風險是什麼？為什麼投資報酬率沒有被證明是合理的？
Example three. After asking Chacht to refine a cold outreach email, you tell the AI, "You are now that VP of marketing I'm targeting." You get 50 cold emails like this every day. Read
例子三。在讓 ChatGPT 細化一封冷開發郵件後，你告訴 AI，「你現在是我要針對的那個行銷副總裁。」你每天收到 50 封像這樣的冷郵件。閱讀
the email you just wrote and tell me your immediate unfiltered reaction.
你剛寫的郵件，告訴我你立即的、未經過濾的反應。
Which specific sentences make you hit the delete button and why?
哪些具體的句子讓你點選刪除按鈕，為什麼？
As you can see by now, the red team technique helps us anticipate challenges, giving us the opportunity to strengthen our message before it reaches a realworld audience. Pro tip: be
正如你現在看到的，紅隊技巧幫助我們預測挑戰，讓我們有機會在訊息到達真實世界的受眾之前加強它。專業提示：對
extremely specific with the persona you ask the AI to adopt. Don't just say act as a critic. Give it a detailed role with clear motivations like you are a
你要求 AI 採用的角色要非常具體。不要只說「扮演批評者」。給它一個有明確動機的詳細角色，比如「你是一個
risk averse CTO whose main concern is data security. The more specific the persona, the more insightful the feedback. Pro tip number two, turn the AI's critique into an actionable to-do
風險規避的技術長，主要關心的是資料安全」。角色越具體，反饋越有見地。專業提示二，把 AI 的批評變成可執行的待辦
list. Use another follow-up prompt like based on the weaknesses you just identified, help me rewrite the three weakest sentences in the original email.
清單。使用另一個後續提示，比如「根據你剛剛識別的弱點，幫我重寫原始郵件中最弱的三句話」。
This closes the loop and helps you instantly improve your work based on the feedback. All right, our fourth hack is a technique I call blueprint scaffolding. In a nutshell, this forces
這形成閉環，幫助你根據反饋立即改進你的工作。好的，我們的第四個技巧是我稱之為藍圖搭建的技術。簡而言之，這迫使
the AI to explain its step-by-step reasoning before it delivers the final output. It's like finalizing the blueprint of a house before actually building the house. Diving right into an
AI 在交付最終輸出之前解釋它的逐步推理。這就像在真正建造房子之前先最終確定房子的藍圖。直接進入一個
example, I start off with a basic prompt like, I offer an online course called the Workspace Academy. I need a marketing campaign brief for the Q4 holiday promotion. And the result is
例子，我從一個基本提示開始，比如「我提供一個叫做 Workspace Academy 的線上課程。我需要一個 Q4 節日促銷的行銷活動簡報。」結果
extremely generic here uh and includes a lot of information. I don't really need for example like tracking UTM's measurement operations and roles risks and mitigation right I don't need to see
非常通用，而且包含很多我不太需要的資訊，例如追蹤 UTM、測量操作和角色、風險和緩解措施，對吧，我現在不需要看到
any of this right now with blueprint scaffolding the improved prompt looks something like this I offer an online course called the workspace academy blah blah blah this is all the same and then
這些。有了藍圖搭建，改進的提示看起來像這樣：「我提供一個叫做 Workspace Academy 的線上課程等等等等，這些都是一樣的」，然後
first outline the standard sections of a professional brief and give me a one-s sentence description for each section by seeing this blueprint first I immediately realized okay it's giving me
「首先概述專業簡報的標準部分，並為每個部分給我一句話的描述」。透過先看到這個藍圖，我立即意識到，好的，它給了我
way too much information. I don't need 18 bullet points or sections, right? So, I can course correct immediately.
太多資訊。我不需要 18 個要點或部分，對吧？所以，我可以立即修正方向。
There's too much irrelevant information. Apply the 80/20 rule and give me only the essential sections for an email marketing campaign with a three email sequence. All right. And this tightened
「這裡有太多不相關的資訊。應用 80/20 法則，只給我一個有三封郵件序列的電子郵件行銷活動的必要部分。」好的。這個精簡後的
up version is way better. So, I just end with, okay, let's just remove seven to eight cuz those again are not that important. And then proceed with fleshing out the entire brief. Right?
版本好多了。所以，我只是最後說，「好的，我們把七到八刪掉，因為那些再次不是那麼重要。然後繼續展開整個簡報。」對吧？
And this final output is way more targeted and relevant than the generic response we received earlier. The rule of thumb here is for complex tasks that require multiple steps, nuance, or a
這個最終輸出比我們之前收到的通用回應更有針對性和相關性。這裡的經驗法則是，對於需要多個步驟、細微差別或
specific structure, always ask to first break down its thought process, review and make changes as needed, then execute. It's like reviewing an architect's blueprint before they start building. Right? by spotting an
特定結構的複雜任務，總是先要求分解其思考過程，根據需要審查和修改，然後執行。這就像在他們開始建造之前審查建築師的藍圖。對吧？透過在他們澆混凝土之前發現
incorrect measurement before they pour the concrete. We're safe from having to tear down the whole thing later. And if you watch my video on the GPT5 update, asking Chacht to articulate its steps
不正確的測量，我們就不用以後拆掉整個東西了。如果你看過我關於 GPT5 更新的影片，讓 ChatGPT 闡述它的步驟
forces the invisible router to select a more powerful reasoning path, which leads to a more accurate and well ststructured final output. Pro tip, we can take this technique a step further
會迫使隱形路由器選擇一條更強大的推理路徑，從而產生更準確、結構更好的最終輸出。專業提示，我們可以把這個技術更進一步
by defining success metrics upfront. For example, I need a social media campaign brief. First outline the steps and for each step define the success metric. For instance, for our competitor analysis,
，透過預先定義成功指標。例如，「我需要一個社交媒體活動簡報。首先概述步驟，對於每個步驟定義成功指標。例如，對於我們的競爭對手分析，」
the metric is a report with three actionable takeaways. This builds clarity and accountability into the output from the very beginning. If you enjoyed these tips, you should definitely check out my top five chatbt
「指標是一份有三個可執行要點的報告」。這從一開始就在輸出中建立了清晰度和問責制。如果你喜歡這些技巧，你一定要看看我的 ChatGPT
use cases for professionals. See you on the next video. In the meantime, have a great one.
專業人士五大使用案例。下支影片見。在此同時，祝你一切順利。
點擊句子跳轉到對應位置
Diving right in, we have the prompt reversal technique for Chacha BT. And here's what it does. You know that back and forth we all have with AI. After
直接進入主題，我們有 ChatGPT 的提示反轉技巧。它的作用是這樣的。你知道我們和 AI 的那種來回，在
that initial prompt, we get a first result that's maybe 50% of what we want.
初始提示後，我們得到的第一個結果大約是我們想要的 50%。
So, we ask for some changes. The second result gets us to 60% and this goes on and on until we finally land on something that's about 90% of what we
所以，我們要求一些修改。第二個結果讓我們達到 60%，然後這樣持續下去，直到我們最終得到大約 90% 我們
wanted, right? What if we could skip all that back and forth and just jump straight to the near-perfect result every single time? That's what prompt reversal does. Think of it like cooking
想要的結果，對吧？如果我們可以跳過所有那些來回，每次都直接跳到接近完美的結果呢？這就是提示反轉的作用。把它想像成沒有食譜的烹飪
without a recipe. You add some salt, taste it, add some pepper, and keep tweaking until the dish is perfect.
。你加一些鹽，嚐一嚐，加一些鬍椒，不斷調整直到菜餚完美。
Right? Prompt reversal is like having a robot scan that final dish and producing the exact recipe we needed from the start so that next time we nail it in
對吧？提示反轉就像讓一個機器人掃描那道最終的菜，從一開始就產生我們需要的確切食譜，這樣下次我們就能一次就做對
one shot. Let's go through a real world example so you see exactly how this works. I start off by asking Chachi BT to analyze our main competitor, Anthropic, and walk me through their
。讓我們透過一個真實世界的例子，讓你看看這到底是怎麼運作的。我首先請 ChatGPT 分析我們的主要競爭對手 Anthropic，並帶我瞭解他們的
business strategy. I receive a comprehensive overview of Anthropic's entire business. But this is way too dense. I'm not sure what I should focus on. So, I course correct and I say,
商業策略。我收到了 Anthropic 整個業務的全面概述。但這太密集了。我不確定應該專注於什麼。所以，我修正方向說，
"Hey, this is too dense. Restructure this into the SWAT analysis format.
「嘿，這太密集了。把這個重新組織成 SWOT 分析格式。」
Strengths, weaknesses, opportunities, and threats. I want three bullet points per section and use clear and simple words." and I get a targeted SWAT analysis. But now this is too concise.
優勢、劣勢、機會和威脅。我想要每個部分三個要點，使用清晰簡單的詞語。」然後我得到了一個有針對性的 SWOT 分析。但現在這太簡潔了。
The bullet points need more detail. So I refine again. Hey, uh, this is now too concise. Flesh out each bullet point and add a subheading under each section called our strategic response and
要點需要更多細節。所以我再次細化。嘿，呃，這現在太簡潔了。把每個要點展開，並在每個部分下新增一個叫「我們的策略回應」的子標題，並
suggest one concrete action that we can take. Finally, I get exactly what I want. Strengths, three supporting bullet points, recommended action. Weaknesses, three subordinating bullet points, recommended action. Same thing for
建議一個我們可以採取的具體行動。最後，我得到了我想要的。優勢，三個支援的要點，推薦行動。劣勢，三個次要的要點，推薦行動。機會和
opportunities and same thing for threats. Perfect. Now, here's where the magic happens. Instead of accepting this result and moving on, I add one final prompt. Reverse engineer our conversation and write the single prompt
威脅也是同樣的結構。完美。現在，這裡是魔法發生的地方。不是接受這個結果然後繼續，我新增一個最終提示。反向工程我們的對話，寫出那個
that would have produced my final response in one go. And look at this.
能一次產生我最終回應的單一提示。看看這個。
Chatbt outputs the reverse engineered prompt in a code block for easy copy pasting. And we can test this out. I'm just going to copy this code, open up a
ChatGPT 在程式碼塊中輸出反向工程的提示，方便複製貼上。我們可以測試一下。我只要複製這段程式碼，開啟一個
new chat, paste it in, click enter, and I'm just going to fast forward this part here.
新的對話，貼進去，按 Enter，我會快進這部分。
And as you can see, we got that perfect detailed analysis in one step. That's pretty awesome. Not only does this technique save us so much time and
如你所見，我們一步就得到了那個完美詳細的分析。這真的很棒。這個技巧不僅長期為我們節省了大量時間和
effort over the long run, but it also captures all those little details we refined along the way. And it helps us become better prompt writers because we can actually see how optimized prompts
精力，它還捕捉了我們一路上細化的所有小細節。而且它幫助我們成為更好的提示編寫者，因為我們實際上可以看到最佳化的提示是如何
are structured. Pro tip, people often ask me which prompts I save to my prompts database in Notion. And honestly, most of these prompts come from the prompt reversal technique
結構的。專業提示，人們常問我把哪些提示儲存到我在 Notion 的提示資料庫。說實話，這些提示大多來自提示反轉技巧，
because those are the best prompts I use repeatedly. Speaking of writing better prompts, today's sponsor, HubSpot, has an awesome resource I'm pretty impressed with. It's their free ebook for professionals, Supercharge Your Workday
因為那些是我反覆使用的最好的提示。說到寫更好的提示，今天的贊助商 HubSpot 有一個很棒的資源，我印象深刻。這是他們為專業人士準備的免費電子書，《用 ChatGPT 提升你的工作日》
with ChachiBT. Longtime viewers know I made an entire video on Google Gemini's guide for professionals. And while that was a good start, the HubSpot ebook goes way deeper into actionable strategies
。長期觀眾知道我做過一整支關於 Google Gemini 專業人士指南的影片。雖然那是一個好的開始，HubSpot 的電子書更深入地探討了
for specific corporate roles. In fact, their section on prompt databases inspired a popular tip for my corporate workshops, which is to assign one single gatekeeper per team to add, refine, and
針對特定企業角色的可行策略。事實上，他們關於提示資料庫的部分啟發了我企業培訓中的一個熱門技巧，就是為每個團隊指定一個把關者來新增、細化和
remove templates from a shared prompts database. I'll leave a link to the free ebook in the description, and thank you HubSpot for sponsoring this portion of the video. All right, moving on to hack
從共享提示資料庫中移除模板。我會在描述中留下免費電子書的連結，感謝 HubSpot 贊助這部分影片。好的，進入第
number two, a technique I call the fiveinut amplifier. diving right into a real world example. When I was a marketing manager at Google, I had to rely on the product and sales teams to
二個技巧，我稱之為五分鐘放大器。直接進入一個真實世界的例子。當我在 Google 擔任行銷經理時，我必須依賴產品和銷售團隊來
provide content for my campaigns. I'd be like, "Hey, you're going to send me those slides tomorrow, right?" And they'd be like, "Oh, yeah, for sure.
為我的活動提供內容。我會說，「嘿，你明天會把那些幻燈片發給我，對吧？」他們會說，「哦，是的，當然。」
Promise." A week later, I'd be like, "Yo, where are the slides?" And they'd be like, "Oh, I'm so sorry. I was busy pitching clients and generating shareholder value, something you
「保證。」一週後，我會說，「嘿，幻燈片在哪裡？」他們會說，「哦，我很抱歉。我忙著向客戶推銷和創造股東價值，這是你們
marketing folks won't understand." And I'd be like, "Oh, go yourself." Just kidding. I love the sales team.
行銷人員不會理解的事情。」我會說，「哦，去你的。」開玩笑的。我愛銷售團隊。
Anyways, with ChadBt, as long as I can get my hands on their main slide deck, I can amplify that content all by myself.
總之，有了 ChatGPT，只要我能拿到他們的主要幻燈片，我就可以自己放大那些內容。
For example, I can ask the AI to create an engaging 10 question quiz based on their slides with multiple choice options and the correct answer indicated. Boom. Now, we can test the
例如，我可以要求 AI 根據他們的幻燈片建立一個有吸引力的 10 題測驗，有選擇題選項並標明正確答案。嘣。現在，我們可以測試
audience's knowledge and keep them engaged. Next, I'd have Catchup draft an internal recap email for stakeholders who couldn't attend the event, summarizing the key takeaways and product updates. Then, I might ask the
觀眾的知識並保持他們的參與。接下來，我會讓 ChatGPT 為無法參加活動的利益相關者起草一封內部回顧郵件，總結關鍵要點和產品更新。然後，我可能會要求
AI to create an external client-f facing infographic that pulls out the most impactful stats from those original slides. And now, I have a follow-up asset to share with clients. And this
AI 建立一個對外的客戶資訊圖表，從那些原始幻燈片中提取最有影響力的統計資料。現在，我有了一個後續資產可以與客戶分享。而且這
works across any department or role, right? The sales team can turn an industry report from marketing into cold emails for client outreach, a LinkedIn post to generate leads and a list of
適用於任何部門或角色，對吧？銷售團隊可以把行銷的行業報告轉換成客戶開發的冷郵件、產生潛在客戶的 LinkedIn 貼文和
talking points for their next call. HR can turn a 1-hour webinar transcript into a quick reference guide with step-by-step instructions, an FAQ for the company internet, and a knowledge
下次通話的談話要點。人力資源可以把一小時的網路研討會文字稿轉換成帶有步驟說明的快速參考指南、公司內網的常見問題，以及
check quiz for attendees. The benefits should be pretty obvious by now. We basically save hours of manual reformatting and rewriting by using AI to repurpose one higheffort piece of
參加者的知識測驗。好處現在應該很明顯了。我們基本上透過使用 AI 重新利用我們已有的一個高投入內容，節省了數小時的手動重新格式化和重寫
content we already have access to. Pro tip, be selective about your source material and only amplify what I call pillar content, your highquality proven work like a successful presentation or a
。專業提示，對你的源材料要有選擇性，只放大我所說的支柱內容，你高品質的經過驗證的工作，比如成功的簡報或
datari report because the AI will magnify the quality of what you give it.
資料包告，因為 AI 會放大你給它的品質。
Garbage in, garbage out, right? If you find my AI videos helpful, by the way, I'm actually developing an entire course on evergreen skills and universal principles to master any AI platform,
垃圾進、垃圾出，對吧？順便說一下，如果你覺得我的 AI 影片有幫助，我實際上正在開發一整門關於常青技能和通用原則的課程，幫助你掌握任何 AI 平臺，
giving you a futurep proof framework that never becomes obsolete. If you're interested in learning a practical and timeless AI system, click the link below to join the weight list. All right,
給你一個永不過時的未來防護框架。如果你有興趣學習一個實用且永恆的 AI 系統，點選下面的連結加入等待名單。好的，
moving on to strategy number three, the red team technique. This is a simple two-step method. First, you ask the AI to create something from your perspective. Then, and here's the
進入策略三，紅隊技巧。這是一個簡單的兩步方法。首先，你讓 AI 從你的角度建立一些東西。然後，這裡是
important part, you immediately ask it to flip the script and adopt a critical and opposing persona. Let's go through some examples. First up, job applications. You first ask Chashib to
重要的部分，你立即讓它翻轉劇本並採用一個批判性的對立角色。讓我們看一些例子。首先是求職申請。你先讓 ChatGPT
tailor your resume for a job description. Instead of stopping there, you then flip the script by following up with, "Now act as a hiring manager for this role. You're extremely busy and
為一個職位描述客製化你的履歷。不是停在那裡，你然後透過跟進來翻轉劇本，「現在扮演這個角色的招聘經理。你非常忙，只有
only have 60 seconds to scan the resume you just helped me write. What are your immediate red flags?" Example two, after asking AI to draft a business proposal for your CFO, you
60 秒來掃描你剛幫我寫的履歷。你立即的紅旗是什麼？」例子二，在讓 AI 為你的財務長起草一份商業提案後，你
redte team that proposal with you are now the company's chief financial officer CFO and your primary goal is to cut unnecessary costs. Read the proposal you just generated and critique it.
用「你現在是公司的財務長，你的主要目標是削減不必要的成本。閱讀你剛生成的提案並批評它」來對那個提案進行紅隊測試。
What's the biggest financial risk? Why isn't the ROI justified?
最大的財務風險是什麼？為什麼投資報酬率沒有被證明是合理的？
Example three. After asking Chacht to refine a cold outreach email, you tell the AI, "You are now that VP of marketing I'm targeting." You get 50 cold emails like this every day. Read
例子三。在讓 ChatGPT 細化一封冷開發郵件後，你告訴 AI，「你現在是我要針對的那個行銷副總裁。」你每天收到 50 封像這樣的冷郵件。閱讀
the email you just wrote and tell me your immediate unfiltered reaction.
你剛寫的郵件，告訴我你立即的、未經過濾的反應。
Which specific sentences make you hit the delete button and why?
哪些具體的句子讓你點選刪除按鈕，為什麼？
As you can see by now, the red team technique helps us anticipate challenges, giving us the opportunity to strengthen our message before it reaches a realworld audience. Pro tip: be
正如你現在看到的，紅隊技巧幫助我們預測挑戰，讓我們有機會在訊息到達真實世界的受眾之前加強它。專業提示：對
extremely specific with the persona you ask the AI to adopt. Don't just say act as a critic. Give it a detailed role with clear motivations like you are a
你要求 AI 採用的角色要非常具體。不要只說「扮演批評者」。給它一個有明確動機的詳細角色，比如「你是一個
risk averse CTO whose main concern is data security. The more specific the persona, the more insightful the feedback. Pro tip number two, turn the AI's critique into an actionable to-do
風險規避的技術長，主要關心的是資料安全」。角色越具體，反饋越有見地。專業提示二，把 AI 的批評變成可執行的待辦
list. Use another follow-up prompt like based on the weaknesses you just identified, help me rewrite the three weakest sentences in the original email.
清單。使用另一個後續提示，比如「根據你剛剛識別的弱點，幫我重寫原始郵件中最弱的三句話」。
This closes the loop and helps you instantly improve your work based on the feedback. All right, our fourth hack is a technique I call blueprint scaffolding. In a nutshell, this forces
這形成閉環，幫助你根據反饋立即改進你的工作。好的，我們的第四個技巧是我稱之為藍圖搭建的技術。簡而言之，這迫使
the AI to explain its step-by-step reasoning before it delivers the final output. It's like finalizing the blueprint of a house before actually building the house. Diving right into an
AI 在交付最終輸出之前解釋它的逐步推理。這就像在真正建造房子之前先最終確定房子的藍圖。直接進入一個
example, I start off with a basic prompt like, I offer an online course called the Workspace Academy. I need a marketing campaign brief for the Q4 holiday promotion. And the result is
例子，我從一個基本提示開始，比如「我提供一個叫做 Workspace Academy 的線上課程。我需要一個 Q4 節日促銷的行銷活動簡報。」結果
extremely generic here uh and includes a lot of information. I don't really need for example like tracking UTM's measurement operations and roles risks and mitigation right I don't need to see
非常通用，而且包含很多我不太需要的資訊，例如追蹤 UTM、測量操作和角色、風險和緩解措施，對吧，我現在不需要看到
any of this right now with blueprint scaffolding the improved prompt looks something like this I offer an online course called the workspace academy blah blah blah this is all the same and then
這些。有了藍圖搭建，改進的提示看起來像這樣：「我提供一個叫做 Workspace Academy 的線上課程等等等等，這些都是一樣的」，然後
first outline the standard sections of a professional brief and give me a one-s sentence description for each section by seeing this blueprint first I immediately realized okay it's giving me
「首先概述專業簡報的標準部分，並為每個部分給我一句話的描述」。透過先看到這個藍圖，我立即意識到，好的，它給了我
way too much information. I don't need 18 bullet points or sections, right? So, I can course correct immediately.
太多資訊。我不需要 18 個要點或部分，對吧？所以，我可以立即修正方向。
There's too much irrelevant information. Apply the 80/20 rule and give me only the essential sections for an email marketing campaign with a three email sequence. All right. And this tightened
「這裡有太多不相關的資訊。應用 80/20 法則，只給我一個有三封郵件序列的電子郵件行銷活動的必要部分。」好的。這個精簡後的
up version is way better. So, I just end with, okay, let's just remove seven to eight cuz those again are not that important. And then proceed with fleshing out the entire brief. Right?
版本好多了。所以，我只是最後說，「好的，我們把七到八刪掉，因為那些再次不是那麼重要。然後繼續展開整個簡報。」對吧？
And this final output is way more targeted and relevant than the generic response we received earlier. The rule of thumb here is for complex tasks that require multiple steps, nuance, or a
這個最終輸出比我們之前收到的通用回應更有針對性和相關性。這裡的經驗法則是，對於需要多個步驟、細微差別或
specific structure, always ask to first break down its thought process, review and make changes as needed, then execute. It's like reviewing an architect's blueprint before they start building. Right? by spotting an
特定結構的複雜任務，總是先要求分解其思考過程，根據需要審查和修改，然後執行。這就像在他們開始建造之前審查建築師的藍圖。對吧？透過在他們澆混凝土之前發現
incorrect measurement before they pour the concrete. We're safe from having to tear down the whole thing later. And if you watch my video on the GPT5 update, asking Chacht to articulate its steps
不正確的測量，我們就不用以後拆掉整個東西了。如果你看過我關於 GPT5 更新的影片，讓 ChatGPT 闡述它的步驟
forces the invisible router to select a more powerful reasoning path, which leads to a more accurate and well ststructured final output. Pro tip, we can take this technique a step further
會迫使隱形路由器選擇一條更強大的推理路徑，從而產生更準確、結構更好的最終輸出。專業提示，我們可以把這個技術更進一步
by defining success metrics upfront. For example, I need a social media campaign brief. First outline the steps and for each step define the success metric. For instance, for our competitor analysis,
，透過預先定義成功指標。例如，「我需要一個社交媒體活動簡報。首先概述步驟，對於每個步驟定義成功指標。例如，對於我們的競爭對手分析，」
the metric is a report with three actionable takeaways. This builds clarity and accountability into the output from the very beginning. If you enjoyed these tips, you should definitely check out my top five chatbt
「指標是一份有三個可執行要點的報告」。這從一開始就在輸出中建立了清晰度和問責制。如果你喜歡這些技巧，你一定要看看我的 ChatGPT
use cases for professionals. See you on the next video. In the meantime, have a great one.
專業人士五大使用案例。下支影片見。在此同時，祝你一切順利。