How to write a cover letter with AI
AI. That large ball of raw, synthetic intelligence on everyone’s tongue recently. And so so available. Free intelligence for everyone!
But how do you use it? How do you shape it? How do you get that ball of raw intelligence to do what you want, in the way you want, when you want?
Well, it’s an emerging skill called “Prompt Engineering” or “Prompt Design”. It’s the difference between a good output and a great one you can use in place of your own work.
Let’s look at how to use AI to write cover letters. To do this, we’re going to write a GPT-3 prompt that generates professional cover letters.
1. Writing a Good Prompt
Writing a prompt is pretty straight forward. Tell GPT-3 what to do.
Write a professional cover letter.
With this instruction, GPT-3 will write a cover letter. But it won’t be very specific. GPT-3 does better with more instructions.
Tip #1: “Specific instructions get specific results”.
Write a 500-word professional cover letter for a job as a Restaurant Manager.
That’s better, but your output will be very generic.
A good cover letter responds to a specific job listing. GPT-3 can add those special touches, but we have to instruct it.
Tip #2: “Add context to your request to get more relevant output”
Write a 500-word cover professional cover letter for a job as a Restaurant Manager with the following description:
“Responsible for all front-of-the-house (FOH) functions on an opening, mid- or closing shift, including guest relations, supervision of all FOH staff and staffing levels, proper restaurant ambiance, housekeeping, and set-up, food & beverage quality, safety and pace.”
That will produce a better cover letter, but it won’t mention you. To get a personalized one, one that’s about you specifically, we should provide information about you.
Tip #3: “To get specific details in an output, include them in your input”.
Write a 500-word cover professional cover letter for a job as a Restaurant Manager with the following description:
“Responsible for all front-of-the-house (FOH) functions on an opening, mid- or closing shift, including guest relations, supervision of all FOH staff and staffing levels, proper restaurant ambiance, housekeeping, and set-up, food & beverage quality, safety and pace.”Mention [Past Skills] in the cover letter.
In the cover letter, mention the following skills you have:
5 years experience in restaurant industry, strong communication skills, experience with high volume bar sales.
Now you’re getting great tailored results. We can still do more though.
Let’s improve formatting. These are not strict rules, but generally, GPT-3 likes clear formatting. It wants to know what goes where. Think of GPT-3 like a really, really smart auto-complete. So let’s put up some bumpers so it’s guaranteed behave the way we want.
Part of this will be reminding GPT-3 at the end of the prompt to write the cover letters.
Tip #4: GPT-3 behaves like an auto-complete. It will continue the structure it’s given.
Write a 500-word cover professional cover letter that addresses the following requirements.
Job: [Job Title]
Job description: [Job Description]
Skills to mention in the cover letter: [Your skills]
COVER LETTER:
*Chefs kiss*
You now have a prompt that reliably writes great cover letters.
To be clear, the previous prompt would work without this added formatting, but hey we’re prompt engineers. Let us ‘gineer that prompt a little.
Prompt Frames (or how to run the same prompt repeatably)
Now what if you want to run this cover letter prompt 25 times a day with slightly different details? And you want to let your equally unemployed friends use it?
The solution is to turn your GPT-3 prompt into something like this:
Now even the guy crashing on your couch can apply to 25 jobs a day. It’s pretty simple too.
By now, certain parts of our prompt are basically copy-paste boxes. We didn’t write a prompt so much as a prompt frame — a structure for a high quality prompt that people can plug their own inputs into.
We can turn our prompt frames into user-friendly forms with Pickaxe, a platform for prompt frames.
Turn it into a Pickaxe
Pickaxe lets you turns parts of your prompt into variables, and turn your whole prompt into a user-friendly form with text boxes so people can run your prompt with their specifics.
Here’s our prompt from earlier.
Write a 500-word cover professional cover letter that addresses the following requirements.
Job: [Job Title]
Job description: [Job Description]
Skills to mention in the cover letter: [Your skills]
COVER LETTER:
On Pickaxe prompt becomes:
Each blue User Input tag is basically a box that can be filled with any input. People using the Pickaxe will fill these User Inputs by typing into text input fields on a form.
Writing your prompt in Pickaxe, simultaneously creates a form where people can interact with it. The form is the front-end to the prompt. We can customize it to help people understand how to use our prompt.
Here’s what the form for our cover letter prompt will look like:
A few minutes later we have a nice, neat, beautiful Pickaxe. Anyone can use your prompt with ease ad speed.
We can even monitor responses, change the underlying model, and rewrite the prompt at anytime.
For more prompt engineering content, watch this space. We’ll be exploring many more prompts.
