Have you ever had an experience like this?
When you talk to an AI, you can speak clearly, but the answer you get is always less than ideal. For example, if you want to write an essay on an environmental topic, you might ask this:
"Help me write an article on garbage sorting."
The AI does give you an article, but it's likely to look like the one on the left in the image below:
(A bland science text on the basics of garbage sorting, generic but lacking in character)
But if you ask the question in a different way:
Background: I am preparing materials for a community environmental month campaign and need to write an advocacy piece that residents can relate to.
Task: Write an article from the perspective of "The Trash Can Monologues" that resonates with residents.
Requirements: The tone should be humorous, incorporating scenes from residents' daily lives and pointing out the real harm of misclassification, about 800 words.
Response: Please list the framework of the article first for my confirmation.
The result obtained is something like the figure below.
Seeing such a big difference, you might be thinking: isn't it just asking AI to write an article, why is it so elaborate? But don't jump to conclusions yet. Next, I'll use three scenarios that you must have experienced to show you the difference between "will ask" and "won't ask".
Scenario 1: Anxiety about copywriting in the circle of friends
Familiar with this situation? Rushing to send a friend's note in time for the holidays, rehashing and changing it ten times, and still feeling dry...
- General Question: Help me write a Chinese New Year friend circle.
- BRTR method:
B: I am an ordinary office worker, usually seldom send friends circle, want to send one on New Year's Eve.
T: Please help me write a circle of friends that won't be too high-profile but still conveys New Year's wishes.
R: To show concern for family and friends, the tone should be warm and natural, not too literary.
R: I need to be given a choice of 3 different style versions.
Scenario 2: Work Email Dilemma
Every time I write an email to my leader, I struggle for half a day with the fear of using the wrong words...
- General Q: Help me write a vacation email.
- BRTR method:
B: I'm a product manager in my company and I need to take a half day off next Wednesday.
T: Need a leave of absence email that is both polite and concise.
R: The reason for the leave (medical examination) needs to be stated, as well as the work handover arrangements.
R: Tone of voice should be professional and appropriate and express willingness to supplement work after the fact.
Scenario 3: Resume Revision Difficulties
Interview season is fast approaching, and looking at resumes but not knowing where to start...
- General question: help me revise my resume.
- BRTR method:
B: I'm a front-end developer with 3 years of experience who wants to make the jump to a big factory.
T: Please help me optimize the description of my project experience in my resume.
R: Need to highlight technical difficulties and solutions, and use data to reflect the value of the project.
R: Please analyze the existing described problems line by line and give suggestions for optimization.
Notice? With the BRTR framework, the AI's answers become more focused. It's like you're talking to a seasoned senior - instead of just saying "help me", you explain your situation (Background), what you want to achieve (Task), what you're specifically asking for (Requirements), and finally what kind of help you're expecting (Response).
In fact, the BRTR framework is essentially a way of efficiently communicating between humans, transformed into a template for conversations with AI. When you master this framework, it's like equipping AI with a "mind-reading" plug-in.
[Disassembles BRTR frame]
Remember, this is not a rigid template, but a thought process. Or take this example of an exchange with a senior predecessor:
1.Background
"Who am I? In what scenario do I need help?"
- Misconstruction: Just say "Write a PowerPoint for me".
- Correctly modeled, "I'm a new product manager and I'm giving a product feature description to the operations team next week."
2.Task
"What specifically do I want to accomplish? What is achieved?"
- Demonstration of error: "Write more vividly."
- Correct modeling: "Need a PPT that allows operations colleagues to quickly understand the highlights of a new feature."
3,Requirements
"What are the specific limitations and expectations?"
- Wrong Demonstration: "Make it look good"
- Correct demonstration:"
- Keep it under 10 pages
- Illustrated with more data and user scenarios
- Needs to be color-coordinated to match company VI (primary color #4A90E2)"
4.Response
"How do I want the AI to respond to me?"
- Wrong demonstration: do not explain, just wait for the result
- Correct model: "Please give me an outline first for confirmation, and then design each page in detail after confirmation"
[Practical tips]
It's not enough to memorize what BRTR means, here are a few tips to make the most of it:
(1)) The Details Make the Difference
Compare these two Backgrounds:
- Common: "I'm going to do a product analysis"
- Enhanced version: "I'm a data analyst for an educational APP, responsible for analyzing user learning behavior data"
(2) Rounding up to zero
Writing an essay, for example, can start with:
Task: "Please start by framing the article."
Next step: "Okay, now please expand the first part based on this framework..."
(3) Leave room for manoeuvre
In Requirements, state "If you feel my requirements need to be optimized, please point them out for discussion first"
[Common Misconceptions]
- The background is too simple.
- Lack of specificity in the description of tasks
- Requirements are too subjective
- Forgot to specify the expected response
[Getting Started: You Try It]
After seeing so many examples, it's your turn to practice. Here's a scenario you must have encountered before:
You want to cook a nice meal for your family over the weekend, but your cooking skills are limited. What would you ask the AI?
- General question: "Teach me to make red meat."
- Ask with the BRTR framework?
Compare your question to this version below:
Background : I am a novice cook and wanted to cook a meal for my family. I have basic cooking utensils at home such as rice cooker and pans.
Task: Please teach me to make a braised pork that even a novice can master.
Requirements: Steps should be detailed, preferably with tips for each step, need to mark the fire and time, explain common mistakes and avoidance methods
Response: Please list the required ingredients and tools first and confirm before detailing the steps.
OK, quality questions are often answered beyond expectations.
[Written at the end]
In fact, the BRTR framework reveals a simple truth: even more powerful AI needs the right "communication style". It's like communicating with an experienced mentor, instead of saying "help me solve this problem", you need to make the situation clear.
Remember this tip: Every time you ask an AI, ask yourself first:
- Have I made my situation and goals clear?
- Are my specific needs clearly expressed?
- What form of help would I like to receive?
Mastering BRTR is like getting a key to unlocking the treasures of AI. It's not just a cue word framework, it's a mindset upgrade.
Whether it's a work presentation, asking for help, or having a conversation with an AI, explaining the ins and outs is always the first step in communication.
