Introduction
Everybody and their mother is talking about AI these days. Will it take our jobs? Honestly, I say: great! Let AI take all the boring stuff I don’t want to do anyway.
Now, before you think I’ve lost my mind, let me clarify: I’m a co-owner of an e-commerce store. And, like most small business owners, my day-to-day is filled with tasks I’d rather not touch—but someone has to do them. I can either hire people (which we currently don’t have the budget for) or grind through them myself. Spoiler: I’ve been wearing my big-boy pants and doing the grinding.
But here’s the thing: I don’t want to grind forever. My dream is to grow this company into a giant, profit-churning machine. Until that day, I’ve got two options:
- Accept that these tasks are my life until we hit the big leagues.
- Automate the living hell out of my company.
Guess which one I’m picking?
That’s right—automation all the way. Either I’ll fully automate most of the boring stuff I do now, or at least partially automate it so I can finish faster or hand it off to someone else without a second thought.
So, in the spirit of New Year’s resolutions, I’ve decided that 2025 will be the year I automate everything that makes sense to automate. And I’m documenting the whole journey for the internet because… why not?
Here’s the thing I’ve learned after using AI for about a year in my business: everything starts with a prompt. If you don’t know how to talk to AI, you’ll struggle to get meaningful results. That’s why step one of my automation journey is all about mastering the art of prompting. Let’s kick things off by demystifying how to talk to AI and building a rock-solid foundation for everything that comes next.
What Is a Prompt, and Why Does It Matter?
The prompt is basically what you tell the AI—simple as that. Open a ChatGPT window, type something into the chat box, and boom: that’s your prompt.
But why does it matter? Think of it this way: imagine the AI is a real person, and you need something from them. What do you do? You explain what you want. Clear communication is key.
Here’s the catch: AI isn’t a person—it’s a machine. It doesn’t inherently understand nuance or context unless you explicitly provide it. The better you communicate, the better your results. The more context you give about the task, the higher the quality of the output you’ll get.
It’s a simple concept: the quality of what you put in directly impacts the quality of what you get out.
Tips for Writing Better Prompts
To get the best results, there are a few key things to keep in mind:
- Be Specific About What You Want
Start by defining your goal. I like to call this the “done state.” What does the finished product look like? For example, if you’re asking the AI to write an email to a client, your done state is a ready-to-send email that you can copy and paste. - Provide Context
The AI needs to know the “who” and “why” behind your request. Who is the email for? Why are you writing it? The more details you provide, the more tailored the response will be. - Give It a Persona
AI works better when you tell it who it’s acting as. In the email example, you might ask it to act as you. Tell it who you are, your tone, and anything else relevant to your “voice.” - Specify the Format
Want a list? A table? A formal email? Tell the AI exactly how you want the information presented. The clearer your instructions, the more accurate the output.
When you provide all these details—your goal, context, persona, and format—the AI knows what you expect, and the chances of getting a great output increase dramatically. Without this information, your results are a gamble: sometimes good, sometimes way off.
The Building Blocks of a Good Prompt
One thing upfront: I don’t think the order of these steps makes a big difference. Depending on your thought process, you can adapt the framework and go with what works best for you. That said, here are the key elements to building a solid prompt:
Give it a Persona
I like to start by giving the AI a persona. Think of it as an actor, and you’re the director telling it what role to play. For example, you could say:
- “You are the General Manager of a Fortune 500 company.”
- “You are a Marketing Content Creator specializing in email campaigns.”
- “You are a Chef crafting a menu for a Michelin-starred restaurant.”
By defining a role, you set the stage for the AI to approach the task with the right perspective.
Be Specific About What You Want
Now, tell it exactly what you want it to do. Don’t just say, “Write an email.” Be clear about what kind of email and how it should look.
For example:
- “Write an email inviting [recipient] to a meeting on January 20, 2025, at 4:00 PM.”
The “done state” is clear: a written email with a suggested meeting.
Provide Context
Once the AI knows its role and the task, give it some background.
For example:
- Who is the email for—a colleague, a client, or a potential partner?
- Why are you sending this email? Is it to schedule a meeting, close a deal, or introduce yourself?
- How formal or casual should it sound?
Context is the glue that holds the task together, ensuring the AI’s response is relevant and aligned with your expectations.
Specify the Format
How should the output look?
- Do you want a simple block of text?
- A list of meeting date options?
- A table summarizing key points?
For example, you could say:
- “Create a bullet-point summary of the last meeting’s discussion.”
- “Include a table of suggested dates and times for the meeting.”
Reference Existing Work
This step isn’t always necessary, but it can be a game-changer when it is.
- If you have a reference email, paste it into the prompt. For example: “Here’s a sample email I like. Format the new email similarly.”
- If you’re creating something like a product description, you could share a well-written description as a guide.
References give the AI a benchmark to emulate, which can dramatically improve the quality and consistency of the output.
Structure Your Prompt Clearly
Lastly, structure your prompt so it’s easy to read and understand—both for the AI and for you if you come back to it later.
- Use Markdown or a simple format with headings and sections.
- Clearly separate references or context from instructions.
Why does this matter? A blob of unstructured text is confusing for the AI and frustrating for you to revisit. Clean formatting makes prompts reusable and easier to tweak.
With these building blocks, you can craft prompts that set the AI up for success, giving you better and more consistent results. A little extra effort upfront saves a lot of time (and frustration) later on.
Iterating for Success
Now, for my absolute favorite part: iteration.
Some modern large language models (LLMs) have pretty big context windows, which means they can “remember” a lot more information within a single conversation. One of the best ways to use this is for iteration—refining outputs step by step until they’re just right.
Let’s say you’ve crafted your nearly perfect prompt, but something is still missing in the output. What should you do? You could go back and tweak the original prompt and try again—especially if you’re planning to save it for future use. However, my go-to approach is to iterate directly within the same chat.
What Is Iteration?
Iteration is simply the process of asking the AI to refine or adapt its output based on your feedback.
- Did it write an email, but leave out a key detail? Ask it to add that detail.
- Want the tone to be more professional—or maybe throw in a joke? Tell the AI to rewrite it with those changes.
- Need it formatted differently? Ask for a bullet-point summary, a table, or whatever you need.
The key here is to keep the chat window open and use the same context. You can go through this process as many times as needed until the output meets your expectations.
Get the AI to Teach You
If you had to iterate a lot, here’s a pro tip: ask the AI to generate a “perfect prompt” based on the final version of the output.
- This gives you a better idea of what you could have done differently in the original prompt.
- You can also save this AI-suggested prompt for future use.
Let the AI teach you how to better talk to it—it’s like a built-in feedback loop!
A Word of Caution: Always Evaluate
Never send out AI-generated content—emails, reports, or anything else—without reviewing it first.
- Always fact-check the details. AIs can hallucinate or add incorrect information.
- Carefully review the tone and content to ensure it matches your expectations.
- Triple-check everything before hitting “send.”
Iteration is powerful, but human oversight is critical to make sure the output is accurate and aligns with your goals.
By iterating within the same context and refining as you go, you can take AI-generated content from good to great—and, in the process, become even better at crafting effective prompts for next time.
Conclusion
How does all this help in automating myself out of a job? Honestly, not that much—yet. I still have to do a lot myself. But it does speed me up, and more importantly, it’s the foundation for everything that follows. Without mastering the prompt, the rest of the journey wouldn’t be possible.
Now that the first step is done, it’s time to take a specific part of my job and figure out how to make it as fast and organized as possible. Stay tuned—I’ll be documenting that process here, too.
If you haven’t already, try this out for yourself. It might seem overwhelming at first, but it’s really not. I don’t even think about all the steps anymore—I just keep the basics in mind, write my prompt, and iterate as needed. Whether it’s for repeatable tasks or one-off content, the key is to start and refine as you go.
As with everything, if the basics are sound, the rest becomes much easier. So go out and test, test, test. The age of AI is here to stay. Jump on the train now, and you’ll be ahead of the game. Wait too long, and you might get left behind.
See you next week (hopefully)!
Leave a Reply