1. Finding your next big idea
Hello again!
2. Spark
Have you ever had a brilliant idea while taking a shower or woken up in the middle of the night with a new business venture you never explored?
3. Spark
Today, we’re constructing the scaffolding to inspire your next startup idea. Thanks to Matt Shumer, CEO of HyperWrite, for inspiring this prompt.
4. Finding your passion
Now, startups might not be for everyone, but everyone certainly has a passion. Whether it be fine art, fly fishing, or Formula 1, we’ll explore how to go from idea to real business and pitch your proposition with ease.
We’re taking ChatGPT’s capability to the next level. Let’s construct the prompt.
5. Assigning a role
Assigning roles to ChatGPT can be a really useful tool for tapping into individual expertise. As we saw in the previous lesson, when negotiating a raise, it’s like having a conversation with two (or more) smart minds.
We’ll apply this to ChatGPT when formulating step-by-step instructions to go from idea to business.
6. Assigning a role
We can ask ChatGPT to take the role of a veteran entrepreneur, inspired by greats like Steve Jobs and Elon Musk. Feel free to change these names to individuals who inspire you.
P.S. Remember to include those XML tags to help ChatGPT understand our prompt’s structure.
7. Idea
After assigning the primary role, enter your free-form thoughts about the idea you want to explore. Remember: a well-defined prompt includes the relevant context to reduce ambiguity.
8. Task
Now, it’s time to highlight the task. Here, ChatGPT acts as a smart analyst to review your idea and propose a product or service with a plan to generate your first dollar in revenue.
9. Response structure
Next comes the response structure. This is the largest part of the prompt that contains smaller sub-sections, breaking down the step-by-step process to develop your idea into something people want.
10. Using few-shot prompting
A useful idea from our prompting toolkit that we can use here is few-shot prompting. Few-shot prompting arms ChatGPT with multiple examples to give the model necessary context. Let’s apply few-shot prompting to the customer acquisition section of the response structure.
11. Using few-shot prompting
I like to highlight non-obvious examples—unique knowledge that the model might not primarily consider in its output. Have a go at using ‘few-shot prompting’ across each of the sub-sections inside the response structure.
12. Getting ChatGPT to think
Research from Stanford and Johns Hopkins University shows that giving language models time to think through responses leads to better performance. We can use thinking XML tags to see ChatGPT’s reasoning and get a more detailed output.
Let’s apply this to our prompt above in the task tag.
13. Getting ChatGPT to think
We’ll now reinforce this inside each sub-section of the response structure. This is what it looks like in the product idea section. You’ll notice there’s nothing inside the thinking XML tags. This is because ChatGPT will use this space to reason through it’s answer when it responds.
Let’s see what ChatGPT replies with.
14. Our big idea
Wow, what a response. Giving time to think and reason through each step has equipped us with a thoughtful plan to build our MVP and make our first dollar from our subscription cupcake service! Try this with something you’re passionate about and see the capabilities of ChatGPT wake up to a new dimension.
15. Let's practice!
It’s time to dive into the exercises where you’ll put your newfound knowledge to the test.