How much time do you spend on prompt structure? Mitchell takes us through a quick demo of the Aisle Prompt Builder, which takes a natural language task description and fleshes it out into the Aisle preferred prompt structure, complete with variables and output structure.
Video transcript
Hi, I'm Mitch at Aisle. Recently, I posted a tutorial video around how you build and manage prompts in Aisle. And some of the comments that I received about that video were around prompting best practices. What is the best kind of prompt structure to use? This is something that comes up and goes away in the dialogue around prompts and AI every so often.
One of the things that we built into Aisle is a prompt builder. we have a structure that we like to adhere to that we think works best. And so we've actually built the capability to describe the kind of prompt that you're looking for and it will automatically map something to that structure. So if you come into and you clicked prompt builder, you can just describe what you're looking to do in natural language. For example, I need a prompt that performs market research on a company.
I will feed a single company name something like that, right? It will think and it will start to generate a prompt. And here you go. Company market research analyzer. You have here a fully structured prompt that you know has the system level stuff.
It has a task overview, talks about inputs, already puts in variables, formats the output, puts constraints in there, and then in dialogue, you can go back and forth and query these assumptions that the prompt builder has made. It also highlights some different things that you might want to change. Once you're happy with it, you can go create prompt and it will put this prompt into our prompt builder and you can you can go from there. Awesome. Thanks,