
In this video I demonstrate perplexity, the best LLM natural language platforms for internet queries I have come cross. Referral Link: perplexity.ai/pro?referral_code=QL63DMDQ
--- type: transcript date: 2023-08-04 youtube_id: HQnb6eAukss --- # Transcript: Perplexity AI: The Answer To Any Question? foreign [Music] which is a platform that accepts natural language as queries and then it scours the internet for potential information you might need and then finally responds back with natural language to your query so as you see here I asked the question of give me the top five Financial stories today and very quickly it gives me both these sources it gives me the answers and then it also gives me related follow-up questions to ask based on that information so the thing that is obviously unique with this is when you compare this to something like Claude or chat GPT without plugins it doesn't give you that ability out of the gate to go and scour the internet for sources and even when you do have something like chat GPT set up with its web browsing capability it isn't too refined and I found it to be more of a struggle than just Googling to be frank so this is an implementation that takes that sort of synthesis between querying the web and natural language and leveraging an llm and does a phenomenal job at it in my opinion now the thing with this is it does also give you the ability to use gpd4 but the thing to note with this is the gpd4 feature and the copilot feature which I'll dive into is something that you'll get a lot more queries with when you do subscribe to the pro tier so I am a pro member of perplexity and I will by the end of the video do a bit of a comparison between perplexity Ai and chat GPT plus and you can sort of decide on what might potentially be worth the money if you're looking to to subscribe to one of these so the next thing to note with perplexed AI is they did recently release a file upload feature so I'll just demonstrate that here so if I go ahead and go over to my downloads here and I just drag over a 10q from Apple's recent recently released quarterly report and I just say how did Apple do this quarter so as soon as you put in that query the thing that's remarkable that I find with this is the speed at which it processes both PDFs as well as the speed to which it queries the internet and gives you that that result back so they've really ironed out something interesting on the back end to make this happen and they've done like I said an amazing job at this so as you can see here based on the file it searched through that and it gave me some relevant details of Apple's report and I could say let's just say what were iPad sales so the nice thing with this is if you're a subscriber to this channel you might have seen us try to build out applications where we query and ask questions of PDFs or pieces of text and the thing with that is it does require a little bit of setup right whereas this it just gives you that ability to drag over your file and just quickly chat with it so it's a different story if you want to build out something like this for yourself or a company or a product idea you might have but if you just want to be able to chat with a PDF or a piece of text throwing it into something like perplexity or Claude which has this feature as well is a really easy way to do that and extract some necessary information that you might need so the other thing with perplexity that it offers is it does give you the ability to share your link if you're looking to share your link and then the nice thing with that shared link is someone else could query and ask follow-up questions based on what you've shared without it actually impacting your first shared uh results so it does give you that ability if someone sees it and they they have an idea of okay what if I ask this follow-up question it gives you that ability without affecting that first share that you have so copilot is a really interesting feature where now if I just enter a new chat which you can very easily with just command K which makes this really easy just to quickly ask a handful of things is if I have co-pilot on and I say give me the latest news so I'll just use some of these similar examples just to demonstrate that sort of real-time information gathering that it has you see it considers sources like we had and we just start to see some of the most recent news that is you know across these different news sources so the nice thing as you might imagine with the the annotations here you can have them embedded within the natural language and then you'll have the responses and you can see it's just going through and giving the information pretty quickly right so the next thing that I just wanted to touch on one more time is that command K feature the one thing I noticed with this when I was playing around with this is you can ask a lot of different questions very quickly so I could say tell me about what's new in Wang chain and if I was to say tell me about what's new with open AI so you can have a number of different threads and it also gives you the ability of say if you have different tabs open here you can have those chat histories running in parallel so it seems subtle but the difference with this is when you compare it to something like chat GPT is if you try to query to two things at once and let's say if I have two tabs open of chat GPT if I just duplicate this and I just click some of their questions here and then try to do it in the second tab you get only one message is available at a time so say if you're a power user of this and you want to just quickly query a number of different things and sort of have a dashboard you could imagine you know having a couple of these parallel to one another it doesn't limit you to that that one message at a time and that is also something that I've noticed recently to be a bit of an issue when using chat EBT is even if I ask it to stop so let's just turn off code interpreter for a moment is if I ask it to stop with the button here and then if I just try and ask quickly another question is it doesn't actually stop so it sort of runs through the background or at least thinks it's running and you have to give it a few seconds so I haven't run into those types of bugs with perplexity AI which is very very nice so another thing to note is you do have the focus state where you can just hone in on you know YouTube or Wolfram Alpha or or whatever you know with is within here I don't usually use this I find just the core offering and the default settings to do what I'd like it to do really well and now the other nice thing with this is in the coding context of say if you want to reference a new AI library for instance and you want to come up with some examples you could quickly come in here write a question and have it write you an example and have that recent documentation right so another limitation of a lot of llm implementations is not being able to access those modern libraries so that's been an interesting use case that I've been playing around with so just sort of a side note with perplexity the thing that I that stood out to me when I just took a look at their blog here is they have some really uh notable investors here so uh individuals from you know board members of databricks uh you know the former CEO of GitHub the former CEO of YouTube and then there's also some Angel Investors and most notably that I noticed were you know the CEO of replit as well as Andre carpathy who's worked on Tesla autopilot as well as the core offering of the GPT models at open AI as well so a huge amount of people have clearly put their support behind perplexity and I can really see why just being a user of it over the past week or so so a bit of a housekeeping thing is right now at time of release of the video there is a quote unquote first birthday celebration where you can get a discount of 25 if you subscribe to one year of service so that will bring this 200 down to 150 so if you are committed to using this for the year I jump on that if not I'll see if I can put a referral Link in the description where I think you'll be able to get at least something like ten dollars off for your first month so a couple other things to note is you do have history like I quickly showed here you are able to search your threads which is something that I wish Chad gbt had that ability within their web interface to search that document feature is great where you can upload and query like I demonstrated the one notable difference is when you compare the upload feature to something like Claude it doesn't allow you to upload more than one file right now so that was something that Claude supports where you can upload up to five files so if you want to compare say different filing types like I did in this one example you wouldn't be able to do that example like I showed in my cloud to demonstration so I've become a pro subscriber I intend to continue to subscribe past the free trial there is also a free trial that you can try on the pro tier of if you want to explore their co-pilot feature now the copilot feature I won't dive into it too much here I just encourage you to get going and try this out sign up make an account really simple to start playing around with it now the copilot feature gives you follow-up questions where it will say say if you want to write an article it might say okay what do you want that article to be about what do you want the title to be so it sort of acts as a bit of an assistant almost where it gives you that back and forth of refining your prompts a little bit so a really great feature and just overall very impressed with perplexity AI I really encourage you especially if you are a chat GPT plus member to consider looking at this as an option as well I think a lot of people will find just that synthesis of real-time data with natural language to be incredibly useful so as always if you found this video useful please like comment share and subscribe and otherwise until the next one
Technical content at the intersection of AI and development. Building with AI agents, Claude Code, and modern dev tools - then showing you exactly how it works.
Weekly deep dives on AI agents, coding tools, and building with LLMs - delivered to your inbox.
Free forever. No spam.
Subscribe FreeNew tutorials, open-source projects, and deep dives on coding agents - delivered weekly.