I do a lot of copywriting, and I’m tired of cleaning up the same AI habits:
Roundabout way of getting to the point
Overuse of AI expressions and isms
Robotic formatting and "dramatic" expressions
So I built a skill specifically for investment writing.
By calling this skill, you can strip out these AI copywriting mistakes and produce outputs that actually feel like language you'd use on the desk.
This week, I’m sharing it with you, along with the techniques I used to build it.
Get my skill below to follow along...
How I built it
Show AI what good looks like
“Write professionally” leaves a lot for the model to guess. Giving it examples makes the target much clearer.
The useful technique is to ask AI to study those examples and explain the decisions behind them. How is the argument structured? What earns a place in the headline? How does the writer connect the evidence to the conclusion?
For this project, I used a dataset of around 50 investment-committee decks.
I asked AI to extract the recurring writing choices and turn them into rules it could apply to a new draft.
That gives you something much more useful than asking it to imitate the tone of one document.
You’re getting it to identify what makes the work effective and carry those principles into a different piece of writing.
Give AI explicit rules to work with
Examples are useful for communicating judgment and taste.
You can also give AI a precise standard for the parts you don’t want it guessing about.
In my case, I used principles from ASD-STE100 Simplified Technical English: a framework for making technical documentation clear and easy to understand.
I borrowed guidance on direct wording, consistent terminology and explicit conditions, then combined it with the rules extracted from the investment decks.
This is how the combo works: The technical-writing framework provides clear ruleset foundations .... and the decks show how that foundation should work in an investment context.
You can use the same approach with your own work. Give AI examples of the output you want, then give it the standards that output needs to meet.
Turn those instructions into something reusable
Once you’ve extracted the rules, have AI organize them into a workflow it can follow: what to look for, how to revise it, and what to check afterward.
I packaged mine as a skill with instructions, examples and guidance for different formats.
That means I can reuse the same approach for a memo, research note or deck without explaining my
