by the end of today's video, you're going to have a complete understanding of how to start leveraging perplexity AI in your workflow. We're going to learn about different features such as Focus, why there even is a discover tab on a chatbot platform, being able to leverage different types of AI in the platform, and so much more. My goal of today's video is to show you everything you can do with perplexity AI and its platform so you can make a conscious decision whether you want to proceed with this kind of AI language model in the context of your workflow. As we are learning in this modern-day age, it seems like everyone kind of wants to choose a platform, whether it's ChatGPT, Anthropic Bard, or Gemini. Now, everyone just kind of goes with whatever they are most comfortable with. So in today's video, let's go ahead and jump into everything we need to know about perplexity AI. First thing I'll note is that this platform is free, and then they have a paid version, so we're going to see both versions here, the free

version allows you to do a lot of search a lot of chats. Seems like the paid version allows us to access higher models and higher usage caps. Let's just go ahead and get a borderline understanding of what this platform is about. The way I want you to think about this platform is a lot more lasered in for very specific use cases. Right off the bat, as you see here, we already have a focus feature. This isn't available in other chatbots. This allows us to laser in the prompt they're about to put in depending on our context, whether that be YouTube, we're looking for a YouTube video, Wolf RME Alpha trying to do some math, academic Reddit, and so on. Therefore, your next question might be, why is there a Discover tab? The purpose of the Discover tab is this is supposed to be like a search engine. So, perplexities AI value point is it can be used like other models like Chad GPT, but it seems like its main selling point here is that it should be used like a new Google. Therefore, if I type in, How do I notice?

how or notice how it was auto-completing by search queries kind of like we see with other traditional search engines or the internet. Maybe experiencing a new wave of search categorization and data manipulation. If we're going to use this like a search engine, let's go ahead and ask a question we would traditionally ask a search engine. So we're going to say, How are volcanoes formed? This is the user interface for these kinds of searches. So what this does, compared to us searching on Google, is it kind of condenses the information so we don't have to search through the sources ourselves and read the sources ourselves. So what it'll do is it'll give an answer and then it'll reference every single data point, that is where they got the information from, as you see from these different article sources. So one would be associated with one, three would be associated with three. If I click these, I can actually go to the source itself. There we go. On top of that, we got our images over here. We can search on

video we can actually generate images based off a search query, and then we get related prompts based off the original search. So, a couple things I want you to take note here. First thing I want you to take note here is notice how there is no ads. Traditional search, we would expect to see ads by now, and on top of that, if we clicked on one of these articles, we would most likely get either a popup for a newsletter or pop up for an ad. First Advantage long term though, they may add ads in this, so I don't know if that's going to stay. Second main point here, notice how when I ask the query, it basically summarizes four different articles into one main answer. This is comparative to if I were to traditionally search it and click through every single article, read every single article, and come to my own consensus. It's expedited the workload in the sense of basically I can know this is probably the most reliable answer due to it being referenced by sources, but how is it identifying what sources are deemed most valuable in the

context of my search query. I'm not too sure. This may be biased, who knows, who knows how they structure their back end when it comes to the sources they choose for the answers. Past all that though, what makes this advantageous is the fact that it's a communication rather than a search. Or I can come down here and say, How do volcanoes erupt? and proceed in a conversation-like flow, relative to me having to research. This research sources and proceed in that manner. Perfect, we understand the search feature, which is probably basically the biggest value point I see out of perplexity AI so far. Now, we must understand the collections feature. We're going to add it to a collection here. We'll say volcano research, hit enter. Now we can reference our chats and categorize them within a collection. So, new collection here. We're going to say volcano report. We can add an icon, we can add a description, and we can even add a prompt to fine-tune further chats within this. First off, let's just create this so you kind understand what it looks like once we

create a collection. We're going to simply come up here to this little square icon. Here, if I go ahead and expand this, it's called The Library. Come over here to your right of volcano report. I'm going to click this. Here is the first major thing we can do and understand about collections. First, notice how the original message that I added to this collection has kind of fine-tuned the suggestions right off the bat. Next, let's actually edit this collection. We're going to go ahead and say format your answers as a bullet point list. So, this is basically any type of new conversation we do within this collection should follow these guidelines. Let's go ahead and click this question here. There we go. Now, every single time I inquire further within this collection for this volcano report, we're going to expect your answers to come out in bullet point form. Obviously, we can do a bunch of other stuff when it comes to how we want to format our outputs. That is a good rule of thumb that you can now understand recruting threads within collections. So, I'm making

this volcano report. But I need to make sure my sources are good, so I'm actually going to laser in here and go to academic to say, Give me a paper that says how Pompeii happened. Really random. I don't even think I spelled Pompeii correct, so here's your answer. I noticed that there is no reference to an academic paper. I'm going to say, Can you give me an academic paper for this? Hit enter. There we go. Actually, there is no specific return of academic paper, so sometimes it won't actually be able to find a paper. Let's actually try again. Again, we'll try something simpler. We say, How do volcanoes work? Enter. There we go. We got our sources for this relevant question. We can go even further here, and we can just basically add these little search by video. It will give us relevant videos. I'm assuming using YouTube's API. Let's go back out from here, though. Check out another feature of flei, which is our ability. If we come over to our settings to adjust the underlying models used within the chat. This is pretty cool.

y'all, so I come down here. As you see, we can actually choose from different models within the market, such as GPT-4, Cloud 2.1, Gemini Pro. Also, we can go ahead and choose image generation models between default and Delay 3. This is important, and this is powerful in the sense that sometimes maybe you want to use GPT for more complex inquiries. Therefore, we can actually switch to it within the settings itself. Another important feature within the settings is our ability to actually turn off the ability to train on our data. So, if that is important for what you're using this model for, make sure to switch that off. Coming over to the Discover tab here, because this is much more like a search engine, they're going to have their own version of a homepage when it comes to relevant articles that got released today. So, if I go, for example, to Nvidia surpass Amazon, we even got a suggestion here of what we could possibly search here. In addition, all these different sources, notice how it takes the topic or

the headline it grabs all the articles that basically exist that are relevant on the internet and then summarize es it by combining all that data into one piece of an output. Furthermore, we can even go as far as ask more questions on the topic. That Just About Does it for plexity AI from my viewership and my audience. I kind of got the vibe that most of us are using this uh the free version for a search engine like experience. If you want to see this compared with Chad gbt or compared with Google search, check out the playlist at the end here as basically I compare the two see which one seems to have better outputs and overall a better experience. Make sure leave a like it's completely free and I'll see you in the next video. What do y'all think are y'all switching to perplex AI or you just going to stick with just Google it that's the play this I was talking about that's a random video that's my face something happens when you click it I'm not too sure but

How To Use Perplexity AI For Beginners

My name is Corbin, an AI developer entrepreneur behind the vision of Webcafe AI. 

Together we will build digital ecosystems. ☕
