How is a prompt used for searches in browsers and AI?

What in the world is a prompt?

Simply, a prompt is what you type into a search bar, ChatGPT or other Large Language Model (LLM) AI that get it to create an output. It’s called a prompt because you’re “prompting” it to create or generate something, whether that’s content, a chart, information, analysis, images or video.
There are two important pillars to abide by when it comes to prompts: context and clarity. The context of the prompt is to provide as much relevant background information as possible so that the AI knows what information to include and, maybe more importantly, what information not to include.
For example, you can upload a document or a series of documents and tell the AI to only reference that information for any responses to your prompt. You can also use old sales scripts you may have and improve or rewrite them based on specific conditions you outline.
That brings us to the second pillar: clarity. You need to provide clear, detailed instructions on what you want the AI to do and how you want it to do it. For example, if you want a certain tone (light, friendly) then tell it that. If you want it to explain pricing and focus on that, be explicit in your prompt.
You also need to keep certain things in mind when writing a prompt. Humans naturally fill in gaps, infer thoughts and can make connections when people mess up speaking, say the wrong word or even the wrong metaphors. We understand context really, really well. That’s a little more difficult for AI, so write your prompt as clearly as possible so that it can’t be misinterpreted easily.

What makes a great prompt

Have AI role play. We’re not joking. It can be valuable to have an AI like ChatGPT or Claude role play for the function you want. For example, you can tell it to imagine it’s a great insurance salesperson and to write some great cold call scripts you can use for your team, then give it the relevant documentation it can draw from. This tends to prime AI, focusing its response on certain parts of its knowledge base rather than others since LLMs like ChatGPT are built to be a lot of different things rather than specific things.
Iterative questioning. Don’t always accept the first response AI gives you. Read it over and ask it to make changes or even ask it why it did certain things. If you read over a piece of information and it’s wrong, tell it. What usually happens is the AI will admit it made a mistake and correct itself. They will also go back to the drawing board and fix or adjust the scripts based on your feedback.
Keep using it. ChatGPT and other AI tools can maintain context within a conversation and learn your preferences over time during extended interactions. The more you use it, the more it learns about how you use it and will get better at giving you responses that are tailored to your style. However, this means you need to keep using it over a medium- or long-term basis. If you just need a set of scripts then you may want to spend a couple days playing around with different ideas first, training the AI on your wants and needs before you get an even better answer later on.
Ask AI to make the prompt. You can also just ask the AI to give you what a great prompt would be. You can either use that or adjust it based on your preferences too. It may feel a bit odd to ask AI to create a prompt to use for AI, but if you’re in a pinch it can be helpful.

Common AI prompt pitfalls

It’s important to remember that you can accidentally do things that make your prompt ineffective. Here are some quick examples:
• Prompts that are too short or too vague.
• Forgetting to add what tone you want.
• Not uploading reference material.
• Giving conflicting instructions.