At first, I treated Perplexity like a nicer and smarter version of Google, but it has snuck into my digital life permanently at this point. I even find myself reaching for it more than my good ol’ Brave browser at times, mainly because it personalizes the results based on whatever prompt I give it. It’s become pivotal to some of my design workflows and even helped me set up my first local LLM. It also pairs really well with note-taking apps like Microsoft Loop and Obsidian.

However, every tool is only as good as the way you use it, especially AI tools - they just reflect your own thinking style back at you. So I started refining my prompts, but using Perplexity goes beyond prompt engineering. I also started poking around the app for hidden gems - it took me an embarrassingly long time to realize it can export chats to your PC. The more I use it, the more I try to incorporate these little tricks I’ve found along the way to get the most out of it. And I regret not doing it sooner…

Using the personalization and customization features

And updating them regularly

If you didn’t already know, Perplexity has a Personalization feature that lets you “introduce yourself”. It functions similarly to a system prompt, where you can tell the AI about who you are, what you expect from it, and anything else to influence how it behaves. I ignored this at first because I’m still used to the time when system prompts were mainly ineffective in AI tools several years ago. But almost every tool’s version of a system prompt is pretty good now. I now always have something in the personalization box, and also update it accordingly.

Spaces, which holds collections of threads in Perplexity, also has its own version of a system prompt. It’s called “custom instructions”, and you’ll find it right at the top of the text box. This lets you give Perplexity specific instructions for those threads, so you can instruct the AI to format its responses in a specific way or change the tone it uses. This is also where you can add links and sites (which Perplexity will prioritize when pulling data).

Using templates

It takes care of the system prompt for you

In the same vein as above, Perplexity’s templates are predefined customizations that instruct the AI how to behave. You can find them by going to Spaces > Templates. And whichever one you pick will fill out the customization box with instructions unique to that template. For example, the Brainstorm Buddy template instructs the AI to “help the user generate and expand on ideas.” This is a great way to quickly prep my threads for specific tasks or topics I want to tackle. There are around 50 templates, I highly recommend checking them out.

Sectioning off my chats by topic

By using Spaces

Speaking of spaces, they’re a fantastic way to compartmentalize the topics you’re researching. I’ll be honest, at first, I usually just grabbed the top thread and prompted the AI regardless of the previous contents of that thread. This actually ended up tainting the results because Perplexity learns from every prompt you give it, which influences how it behaves throughout that specific thread. Now I keep my topics separated, and I use Spaces to do it. Think of threads as files, and Spaces as folders that hold those files. Not only does this give me more relevant results and responses, but it also just looks neater.

Exporting my chats

It ties Perplexity into my local note-taking stack

I recently started incorporating Perplexity into my local note-taking workflow thanks to a feature I’ve overlooked for months - export. Turns out you can export your full threads as PDF, Markdown, and DOCX. This is fantastic for a local plain-text workflow like mine, because all I need to do is export those files straight into my Obsidian vault, and open them in Obsidian. It includes your prompts and all the responses Perplexity gave you, and all of the formatting and clickable links remain intact.

Furthermore, I can export the text and PDF files to a folder synced to my Google Drive, and fetch them directly within NotebookLM. This makes Perplexity one of the best tools to pair with NotebookLM for studying and research. I have a couple of folders dedicated to my Perplexity exports now and routinely export the most useful threads into them. It’s like building a little research stack, except Perplexity does all the fetching for me.

Getting the most out of Perplexity with tiny habits

I actually didn’t heavily rely on Perplexity as a research tool at first, and mostly used it for fun and convenience. But now it’s a core part of many of my projects. And it’s thanks to these little habits that I’ve managed to squeeze more out of it and give myself more a structured user experience.