A four-step method for prompting AI to do what you want: define the role, objective, command, and format. The speaker walks through each step to help you get consistent, usable results from language models.
I'm going to take you from LLM novice to expert using these four steps. Number one role. Give the AI its Persona. If it's a SaaS copywriter with 25 years experience, tell it that it is that. Don't just assume that it's going to be that for you on its own. Number two. State the objective. You are a SAS copywriter with 25 years experience whose job it is to write four pieces of extremely detailed content about our new project Project feature. Number three Specify the command so if you need it to produce a piece of content for your site or a piece of content for pr, give it the individual job, the quantity, and ensure that it knows exactly what the command is. And number four is the format. What format do you want that in? Do you want that to be 800 words? 1500 words? If it's not blog content, do you want it in a spreadsheet? Do you want it in a specific type of format unique to you? If you give it those four commands and then the golden trick is to ask it to give you the prompt that would have got the exact outcome that you're happy with once you are happy with it, and write that back to you and then save that for all your future prompts, you won't have to go through that process and you'll be an LLM master.
