How to Master AI #ai #smartphone #chatgpt #samaltman
How to Master AI #ai #smartphone #chatgpt #samaltman
#PromptEngineering #ChatGPT #WorkAutomation Full video link https://youtu.be/IdyzCTknR98
Ever get frustrated when your AI gives you nonsensical answers? The problem might not be the AI, but your instructions. A 6-year veteran PM at an AI startup with a design background reveals the six core principles for writing prompts like a project plan, turning AI into your brilliant junior colleague—shown in a 'raw' and unfiltered way.
Prompting techniques to make AI your ace assistant (If you don't know this, you'll be at the mercy of AI forever) Communicating with AI, are you still just asking questions? It's time to move beyond instructions and deliver a plan. A 6-year veteran PM at an AI startup reveals everything about designing prompts that turn AI from a simple search tool into an ace assistant that perfectly executes your intent.
The Six Absolute Rules for a Perfect AI Prompt (Practical pro-tips from a 6-year AI startup PM) A good prompt has a formula. Stop asking questions based on gut feeling. A 6-year PM at an AI startup will teach you the six absolute laws for writing fail-proof prompts, verified through thousands of tests. This one video will change your AI.
Timeline: 00:00 Opening: The Real Reason AI Talks Nonsense 01:48 It's Not a Prompt, It's a 'Project Plan' (A PM's Perspective) 04:01 Step 1 (Essential): Task, Context, Example (Just these three will get you decent results) 06:22 Step 2 (Advanced): Persona, Format, Tone (These determine the quality of the output) 08:10 Live Demo: Writing Google Apps Script Code (Before & After) 10:00 Closing: Are You AI's Boss or Its Client?
#ProductivityBoost #AIUtilization #PromptWritingTips #OfficeHacks #LeaveWorkOnTime #WorkEfficiency #ProductManager #PM #Planner #Designer #Marketer
Characteristics of people who can't get AI to work for them lol Are you asking AI vague questions? How high-performers instruct AI to make it a smart assistant. These three things will 5x the quality of your ChatGPT answers
Why Your AI Prompts Are Garbage The AI isn't stupid. Your instructions aren't clear! How to write a prompt 'project plan' to master AI.
Have you ever asked ChatGPT something like, 'Suggest five marketing ideas to attract new customers for our service,' only to be disappointed by generic, useless answers like 'Send newsletters, use social media, influencer marketing...'? lol
Many people expect AI to read their minds perfectly and produce amazing results, but the reality is often a series of irrelevant responses. Why is that? It's not because the AI is dumb, but because we're instructing it like an incompetent boss.
From my perspective as a 6-year PM at an AI startup with a design background, a good prompt isn't a magic spell; it's just like a well-written project plan or requirements document. The clearer the background and objective, and the more specific the desired outcome, the higher the quality.
So in today's video, instead of some vague prompt engineering theory, I'm going to give you six formulas for instructing AI like a high-performer that you can start using today. Once you know this formula, you'll no longer be stressed by AI's nonsense and will be able to manage it like a smart junior colleague sitting right next to you.
Now, the six elements that make up a perfect prompt are: Task, Context, Example, Persona, Format, and Tone. No need to memorize them. Just think, 'How would I explain this task to a new hire?'
Among these, there are three essential elements that form the very backbone of the prompt. They are Task, Context, and Example. Just getting these three right will improve the quality of your results by at least five times.
First, the Task. This is a clear definition of 'what do you want me to do?' The most important thing is to start with a 'verb'. Like 'Summarize,' 'Analyze,' 'Write,' or 'Compare.' Vague, noun-based questions like 'About the marketing plan' will only make the AI wander.
Second, Context. This is the background explanation of 'why does this need to be done?' Without context, the AI can only give you vague answers. Let's take that marketing idea question from earlier as an example. The moment you add context like, 'Our service is a time management app for university students in their twenties, and we've just launched our Instagram channel. Our ad budget is under one million won,' the AI's response becomes much more specific and realistic. You start getting actionable ideas like 'Recruit student supporters' or 'Start an Instagram Reels challenge.'
And third, Examples. This is where you directly show it, 'This is the kind of result I want.' As someone from a design background, I think this is the most powerful cheat code. lol. Showing one good reference is far more effective than a hundred words of explanation. When asking it to write a cover letter for a resume, instead of just asking, show it an example of a successful cover letter you used for another company and say, 'Write it using this structure and style as a reference.' The result will be completely different.
Okay, that covers the essential steps: Task, Context, and Example. With just these three, the AI will operate much more intelligently. Now, let's move on to the advanced stage that will take the quality of your output to the next level: Persona, Format, and Tone.
Fourth, Persona. This is about assigning a role to the AI. Like, 'You are now a copywriter with twenty years of experience,' or 'You are a data analysis expert.' When you assign a persona, the AI uses the specialized knowledge and perspective of that role to generate answers. Instead of just 'Write an ad copy,' saying 'As a 20-year veteran copywriter, write an ad copy for this product' will yield a much more creative and persuasive result.
Fifth, Format. This means clearly specifying the form you want the output in. 'Organize the results in a table format,' 'Write the code in Markdown format,' or 'Create it in the form of an email draft.' If you don't specify this, the AI is likely to just give you a long wall of text. By defining the output format you want beforehand, you can save yourself the trouble of reformatting it later.
Finally, the sixth element, Tone. This is about instructing the mood of the output, the tone and manner. You can give specific instructions like 'in a friendly and witty tone,' 'in a professional and trustworthy tone,' or 'since this is an apology, use the most polite and sincere tone possible.' The same email content can be perceived completely differently by the recipient depending on the tone.
Alright, shall we look at a practical example using these six formulas? Let's assume that as a PM, I need Google Apps Script code that automatically creates Gmail drafts from customer data organized in a Google Sheet.
When you write a specific 'project plan' for it, the AI is no longer a rookie employee talking nonsense, but an ace assistant who perfectly understands my intent and creates the final product.
The key to making this dramatic difference is the backbone of the prompt. I mentioned the six elements that make up a perfect prompt earlier, and among them, there are three essential elements that form the very core.
They are Task, Context, and Example. The AI stands firmly on these three pillars, and the completeness of the output increases by at least fivefold.
The first essential element is Task, a clear instruction for action. In English, it's Task
It's a clear definition of 'So, what do you want me to do?'
Many people ask with vague noun phrases like 'Regarding the marketing plan,' but such requests only make the AI wander.
The most important thing is to start with a verb. As you can see, you must direct the task with a clear 'verb' like 'Summarize this report,' 'Analyze this data,' or 'Write this cover letter.'
The second essential element is Context. In English, it's Context
It's the background explanation of 'why this task needs to be done.' A question without context makes the AI wander. Specific background information determines the quality of the AI's answer.
Remember the vague question from the beginning of the video, 'Recommend some marketing ideas'?
The moment you add specific context like, 'We run a time management app for university students in their 20s, have just opened an Instagram channel, and our budget is less than 1 million won,' the AI's answer becomes much more specific and realistic.
And the third essential element is Example. In English, it's Exemplar
This is what I, as someone with a design background, consider the most powerful cheat code.
Showing one good reference is far more effective than explaining it a hundred times in words. Providing an example similar to the result you want maximizes the AI's understanding.
Okay, that covers the 3 essential elements that form the backbone of a prompt. Just by taking care of these, the AI will operate much more intelligently.
From now on, we'll look at the three advanced elements of an expert's prompt that take the texture of the output to the next level. They are Persona, Format, and Tone.
First is Persona. It's an effective method of assigning a specific role to the AI, making it answer from the perspective of a particular expert. For example, 'You are now a copywriter with 20 years of experience.'
Second is Format. This means clearly specifying the output in a particular form, like a table, Markdown, or an email draft. By defining this in advance, you can get a result that's immediately usable without needing to reprocess it yourself later.
Last is Tone. This is about setting the mood of the output, its tone and manner. Even with the same content, specifying the tone, like 'friendly and witty' or 'in a professional and trustworthy tone,' completely changes how the recipient perceives it.
By using Persona, Format, and Tone like this, you can precisely control the quality and accuracy of the output according to your intent.
Watch the Video
This post is based on our YouTube video. Watch it for more details!
Originally published on YouTube: 10/16/2025