Almost everybody says the same thing about AI prompts. Add more. More detail, more rules, more steps. For a long time, that worked, but not anymore. If you're new here, I'm Dylan. I run an AI consultancy. And lately, a lot of my coaching sessions have been fixing prompts that used to be great, but are no longer useful. The newest models read words more literally. So, the fix is taking words out, not piling more on. I'll show you three words to cut, and also what to use instead. So, let's get into it. Now, there are three kinds or categories of words that we want to cut or alter to get more from the newer AIs today. And the first category of words are going to be rule words, words that are kind of absolutes, meaning that you say always, never, must, etc., trying to get the AI to follow instructions. These actually often degrade the AI's output today because the AI does everything it can to achieve that specific always, sacrificing everything else inside of your skill or your prompt. We'll talk more about why that's the case and how to fix this along with the others later on in this video. That's our first category. Our second category are vague or fuzzy words. When we say something like detailed, professional, summary, etc., words that we have definitions for, but the AI has a definition for that word as well, and it may differ from ours. And that's our second category of words that we need to alter or cut. And then the third one are words when we try to get more effort from the AI. So, when we say think harder, think step by step, whatever else, they used to be a very common tactic people used to use in the past that no longer is relevant and is actually a distraction from any of these models today. So, we need to remove that completely and use a setting instead. So, these are our three main categories. We'll start with the first category, which are rule words. So, like I mentioned, we have absolute terms such as always, never, must, and when you use these words inside of your prompts, inside of a skill or a project, the AI does everything it can at all cost to follow that instruction because the state-of-the-art models today, such as Fable 5, GPT-5.6, whatever else, they follow instructions very well and they're extremely literal. So, when you have too many of these absolute terms inside of your prompts, it will sacrifice everything else to follow that instruction and likely give you a poor output. So, what we want to do instead is we want to have the AI have an outlet so it can take a different path when it needs to. And that's going to be an if-then statement inside of your prompts. So, what does that actually look like? Well, here we have a bad version of the prompt that has many absolutes that it shouldn't. Here we're saying, "Always include the risk section. Never make assumptions. And you must cite sources and for every number." That's the old way of doing things that used to work but no longer works today. An improved version of this is we have a series of outlets for the AI, the if-then statements. So, this new and improved prompt says that if a section has no real risks, then write not found instead of inventing one. Another could be, "If a number isn't in the file I gave you, label it not found. Don't guess." And the last one is if my request is unclear, ask me one question before you start. Each one of these is if-then statements, giving the option for an AI to then take one path or the other, not stopping everything because it didn't follow a certain rule. Quick pause in regular programming. As usual, this video is brought to you by me. Two quick things. First off, below is a 30-day AI Insight series, completely free. You'll get 30 insights in your inbox of how you can apply AI to your business and your work. The second thing is if you'd like to work with me, below are a series of offerings to see if there's a good fit between the two of us. Now, let's get back to the video. And the beautiful part about this category of words as well as the next one is we can use AI to improve them. So, we can have a look at our old instructions for a system prompt inside of a project or a skill, identify the absolute words, and then improve them. And here's a prompt that you can actually steal for yourself. So, say you want to review some of your projects or skills that you use quite often to improve the specific issue. Well, you copy and paste this prompt saying, "Look at the instructions inside of this project or skill." You would swap out each one of these depending on what you're looking at. And you say, "Based on what's in here, are there any absolute words like always, never, must? And if so, I want you to list them back to me. But in addition to doing that, I want you to suggest how we can turn each one of these into if-then rules that fits my actual needs for this specific skill or project. After it's applied those, you can then say, "Okay, now that you've identified these, I'm happy with them. Go ahea