I hope you are already familiar with Perplexity and its basic capabilities. If not, I invite you to my previous post. In today’s post, we will talk about the slightly more advanced capabilities that this tool provides. You have to admit that Perplexity offers a simple and clean interface, making it very easy to use. However, it has a minor drawback: it is a bit more difficult to find advanced options, which are also worth using.
Collections
Probably the most useful functionality, which, however, is not visible enough at first glance, is the collections mechanism What are collections? Well, they allow you to prepare threads within which searches are conducted. Such threads can be easily shared between different users. In addition, they allow you to customize your work with the search engine under specific preferences. For example, if you are a PHP programmer and you care about receiving a specific action plan, then you can create a collection dedicated to such searches.
You can see how to create and use collection in one gif:
If you want to create a collection, then go to the “Library” tab, and then click on “Collection”. A form will appear as below:
The first parameter represents the name of the collection, with which you will later be able to distinguish it from other collections. The emoji and description are supposed to help with this too. 😉
However, the “AI Prompt” field is key. It is optional, but under this inconspicuous name there are great possibilities. With its help you can personalize the search engine’s performance. So that it returns the information most important to you in the format you prefer.
How to do it?
In order to best personalize the search engine’s performance to your liking, put the “AI Prompt” into context. Mention, for example, that you are a Senior PHP Developer and what technology stack you use. This will make the answers you will receive more related to the technologies you use. You’re not likely to get code examples in C++ anymore. The way you solve problems will also be tailored to your skills.
Another important issue is the ability to format your answers. You can say, for example, “return the result in the form of a table with headings X, Y, Z and columns A, B, C.” Alternatively, you can ask that the results be returned in a compact form, with the most important points in bullet-points. You can also ask that a specific step-by-step action plan be returned. This is especially useful if a lot of your queries start with “how to…”. 😉
If you have strongly specified expectations about the returned answers (e.g. you care about JSONs, because you will process them further later), you can also indicate in the “AI Prompt” field some examples. That way there will be a better chance that you will receive responses in a compatible format.
You can use the collection created this way from “Library” > specific collection. When you click on it, you will see all the search threads that have taken place in it. You can also create a new thread.
From now on you can already use your personalized search engine and save time on searches. 🙂
Alright, that was strong, but that’s not the end of Perplexity’s slightly more advanced capabilities, which will surely come in handy in your daily work.
Searching documents (only PRO)
Another useful feature is the ability to upload your documents and search in them for information. So, yes, this means that you can upload a PDF with the documentation of your favorite library or tool and instead of reading it all-search for answers to just what you are interested in/what you need.
ChatGPT also provides similar capabilities, but it’s worth noting two main points that currently (note: this may be changing) differentiate the two tools. Perplexity allows you to not only search for information strictly in the documents provided. The knowledge from them can be juxtaposed with the current content of the Internet, i.e. the news available there. This allows us to check, for example, how up-to-date the uploaded documentation is, or what threads of it have already been discussed in social networks.
On the other hand-Perplexity has a smaller context window, which will make it slightly worse at handling longer documents. It may omit some of the file content from its response.
Focus
When doing a new search in Perplexity, we can select the appropriate “Focus” mode, so the search will be tailored to specific needs.
Among the available options are:
– Web – standard search of the entire Internet,
– Academic – the search engine will focus on academic and scientific searches. It will check academic databases, academic repositories, academic journals and academic publications, among others.
– Math – the Wolfram Alpha service will be used to generate answers, extremely useful for running slightly more complicated calculations.
– Writing – allows you to generate texts based on knowledge derived from the language model. This mode will be better for creative work, relying less on search results.
– Video – searches for information on YouTube. Technically, this happens as Perplexity. reviews and processes transcripts of recordings that may contain the information you are looking for. This happens real time, so no worries-Perplexity will not take care of watching videos while you wait for the query results. 😉
– Social – the search engine will focus on finding information on the Reddit platform. This is especially useful if you want, for example, to know the opinions of Internet users.
– Reasoning (beta) – is a mode for solving complex problems. It uses the OpenAI o1-mini model for its operation. Particularly useful for tasks that require logical thinking and multi-step inference.
Modifiers
It is possible to narrow the field in which the search is conducted in a slightly different way. Modifiers are used for this:
site: – you can indicate the specific site where the searches should be carried out, for example. For example you can type: “How to create component in Angular? site:angular.io”
before: and after: – allows you to narrow the results to a specific time interval. Good example: “Presidential election in USA results after:2024-10-01”
filetype: limits search to the indicated file type, e.g. you can indicate that you want to check PDF files. Example: “Artificial intelligence research site:stanford.edu after:2023-01-01 filetype:pdf”. Yeah, you can combine all this modifiers together.
Summaries
This is not a functionality described extensively by the developers of Perplexity, but it can be very helpful when doing research. Namely, after doing some searches, you can ask for a summary of the conversation so far. This will give you a concise summary of all the issues raised. Its quality is generally so good that you don’t need to change much. In this way, you will create a coherent text about the issues you were looking for information about.
API
Perplexity allows you to integrate the solutions you provide with any system, thanks to API access. This allows you to communicate with the search engine from any other application. Of course, such communication costs extra, but in the PRO plan you get $5 to use every month. This is a powerful solution, but a more detailed discussion of it will come in future posts.
Apps
Perplexity has its own mobile app for iOS and Android, so you can use it in a simpler way on mobile devices. On the other hand, when it comes to working more optimally on desktop devices, it is worth using the extension for Chrome. It makes it possible, among other things, to do searches without having to switch between tabs. Besides, the extension allows you to search for information on the page you are on. Generating summaries from such a page is also a useful feature. I also have good news for macOS users. On October 15, 2024, a dedicated app will be released on the App Store. It promises to be very interesting, but we will see how it will look and work in practice. If you already have any insights or comments-share them please in the comments. 🙂
Pages
Perplexity allows you to generate pages, which later can be easily displayed on the Internet. Especially for you I created this beautiful article:
https://www.perplexity.ai/page/history-of-silicon-valley-inno-sIl2ioXTQfuWs1MopsCfpQ
I find it hard to understand the target group for such a solution. Nowadays, it is not a special achievement to generate a page with some smart-sounding content. Just who is going to read it? Besides, such content has some (but low) value for marketers. Ok, if you want to organize your notes, maybe this is an idea.
Leave a Reply