Your AI's tree of knowledge
When considering what topics to cover, think of your AI's knowledge as being like a tree. 🌳
This analogy can help you structure information effectively, inspire topic areas and spot any gaps. Imagine your knowledge organised into these components of a tree:
The trunk > This is your foundational content and includes:
- Basic information about you and your business
- Personal details you're happy to share
- An overview of how you can help
- Your core focus and approach
- The main services you offer
The branches > These are your core areas of expertise. For example:
- If you're a business coach, your topics might include business growth, strategic planning and team management.
- If you're a life coach, your expertise may cover personal growth, career development and daily habits.
- If you're a fitness coach, you may focus on strength training, exercise routines and nutritional guidance.
The leaves > These cover subtopics, practical scenarios and specific client questions connected with your core topics. For example:
- If you're coaching on team leadership, clients might ask: "How can I improve team communication?"
- If you support clients with their careers, they may say: "I want to change careers - how do I start?"
- If your focus is fitness, you might get questions about specific exercises, such as: "How do I fix my squat form?"
The roots > These are your core values and beliefs that underpin your entire approach and methodology.
These will come through in all your AI training, and you can add specific P&Cs about these. You might also mention some in your coaching methodology instructions.
Building your tree
There isn't a set order, but for a structured approach, build your trunk first, then add one branch at a time, filling out the leaves as you go.
You might have content readily available that you can upload, covering each and every topic. You can also use ChatGPT, or your chosen LLM, to help you quickly create relevant training data.
Prompts to create training data
Here are some prompts you can give to ChatGPT to help you come up with more questions, and then answer them:
Trunk:
“What are 20 questions a [client] might ask a [type] coach?”
Branches & leaves:
“What are 20 questions a [client] might ask a [type] coach about [topic]?”
“What are the 20 most common problems a [target audience member] might have?”
Answering the questions:
"I am the coach with expertise in this topic. Your task is to create answers to all of these questions and display them in a three-column table. The answers should be based on my beliefs about my topic.
First, ask me about my beliefs on my topic, with the goal of identifying my strongest and most unique beliefs. After I provide answers, give me follow-up questions. Then ask me to paste in some content and information that reflects my expertise, worldview and writing style.
When you have sufficient information, provide the answers to all of these questions. Give them in a three-column table, where column 1 is named "prompt" (the question), column 2 is named "completion" (the answer), and column 3 is named "labels". For the label, summarise the topic of the prompt and completion pair in one or two words.
There should be 20 general question and answer pairs and 20 more for each of my topics in the final table. If you cannot complete a question, ask me for more information about my beliefs and approach, or ask me to paste in more content.”