FREE: Llocal Search Perplexity Alternative Ollama RAG Search🤖 Perplexica, Morphic Alternative

What's going on guys Josh pook here and in today's video we are going to be looking at a free open- source alternative to perplexity this tool is called L local search I'm going to show you how you can set it up very quickly within 5 minutes after watching this video and you can use this with any local model any open source model on a llama such as uh quen 2 deep seek llama 3.1 literally anything this one's going to be good

Let's dive right into it [Music] all right guys so here is local L local search I'll leave a link to the repo in the description L loc search is a completely locally running search aggregator using llm agents the user can ask a question and the system will use a chain of llms to find the answer the user can see the progress of the agents and the final answer no open Ai No Google API keys are needed so that's pretty cool you

Don't need any search API Keys any um other like API keys from any llm and you may have seen tools uh if you've been following this channel like this that you do need those things so we've covered tools such as LM answer engine we've covered morphic all right we've also covered perplex okay so I've done deep um reviews on all three of these tools and I even did a comparison I'll leave a link to all those separate videos in the description

Down below if you want to check and see which perplexity alternative is the best fit for you because they all have their different strengths weaknesses Etc right now this one out local search is really good for in terms of uh using the local models all right it's best for using it with local models you're not going to want to use it with Claud or open AI or anything like that okay and another really cool thing is like it's like I just

Said earlier it uses agents and you're going to see what I'm talking about in just a second and I'm not going to go through this whole entire um GitHub repo but basically it is actually probably one of the easiest ones to install all you do is you just get clone this repo so you go get clone get clone and then you would copy that link that I'm going to have in the description once you clone that you're going to CD into

That folder L local search and then this is optional you can create an environment F uh variable I actually didn't do this this is if you're going to change some of the default settings so you could go ahead and look at the code if you want play around with it uh here it is right here well here is the open project um the only thing I really did was okay so I did two things and this one thing I don't know

What maybe maybe it's just me I don't know if it is always like this or maybe it's a bug um but if you at least for me if I didn't do this you would have faced an issue and I'll show you that what to do in just a second so you don't face side issue but the only thing I did was I changed the port to port 332 because I'm already using Port 3000 for a nextjs project right now um that

I'm working on um but right here where the olama host um as for me at least when I get cloned it it didn't have this htttp HTTP coolon for SL slash so I added that there all right if before I added this here I was getting an error okay it was like this is an invalid uh ho host thing right so that could just be because you're meant you could put it in like a environment variable with olama whatever maybe that's

What they intended for but doesn't say that in the GitHub repo and as to at least for me it didn't have that HTTP so make sure you have that there then save it then you can do simply just dog or compose up- D all right so you're just going to run the command then you'll be good to go and then it's going to look something like this well actually when you first open this up it's going to be on light mode

But of course you're probably going to want to put it to dark mode um at least I did now there's a few different things right so this is a I mean they all all the three tools I covered before morphic perlex uh llm answer engine they all kind of have different uis my first impressions on the UI it's not bad it's kind of like At first I was like uh but like I actually do kind of like it you know to

A degree um the font kind of like a claw font so you know um not bad and what you're going to do um so here what does this do this this uh okay oh this is pretty cool so I didn't know this okay so if you press this I icon we're going to do some chats in just a second but you'll see if you press the I icon you can literally see the agent thinking and we're going to dive into that

In just a second but uh up here in the top right is the settings so this is going to autoload your um as long as you put that HTTP like I told you it's going to autoload your old Lama uh models right here so you can select the different ones by default it is news Hermes 2 Pro llama so if so if you use this it will automatically download it and then it's going to show you other models all right so

I'm currently using llama 3.1 but like I said you can use any model just go to ama.com download whatever model you want and then here um you're going to see some other settings here the llm uses Thea system message so you can see that here I'm not going to go over it but you could change that and then the amount of Agents uh agent will be set to a temperature or the agent will be set to a temperature of you can

Change of temperature here the context size for tokens um a maximum of 30 iterations and actions evaluations per uh question so it uh comes with a chroma database a chroma so there's Vector DB um database entries need a minimum of score of 0.5 to be returned to the llm a DB query will return a maximum of 10 results web scraper so I believe it uses sir sir I always say this wrong sirx whatever however you say that and uh we'll return

A maximum of three websites per query so we could go ahead and change this to like what's the maximum uh okay so it can be 1 to 30 we could do 10 um and then text chunks splits so Source text will be split into chunks of x amount character trunking will try to split the text into at new lines or paragraphs the trunk will overlap at whatever right so pretty cool settings this is something that you can't do in even a

Lot of the other well you could do in the other tools some of them aren't as simple to do that like this as that with the whole rag uh configuration but this is something obviously you can't do in perplexity um okay so now let's go ahead and just test like a question so let's just you know ask it how do I make a chocolate cake five star and let's see what we get here so okay so it's interesting it looks like

It's actually using um different search engines like here we have Quant or Quant however you say it then we have Brave we have Google we have Brave so that's kind of interesting um I don't know if any of the other open source ones that we've covered on this channel do that which is kind of cool that it gets it from from different sources um yeah so anyways basic I think we have the uh token output um I don't know if you

Can change that somewhere but uh yeah the output was pretty small pretty short it's not a super in-depth question we could say hey give us a super in-depth question answer it really long but essentially what's pretty cool is that if I go like this as you can see we see that it searched the web and then it um queried the data base and yeah so and let me see oh yeah if we go like this we can even see uh more

So let's see see kind of just a little bit of the thought process here um do how do I make a chocolate cake Five Star Quality so entering chain with 140 tokens so you see agent scratch Pad history system format uh markdown um okay so I think that's the previous yeah this the previous context and then do I need to use a okay so this is actually right here so it's asking itself do I need to use a tool which is

Yes okay cool let's use the web search action all right it's going to search that then tool starts it's you can see what it searches for and then you can see 933 tokens thoughts and it has a lot of different thoughts here this is all base around what it pulled from the internet and then it's going to formulate that into to a answer which is right here okay as you can see I'm not going to read all this to bore you

Guys think none of you really care how to make a chocolate cake but on the left hand side you can see if we click this uh um if we click on the top left there's a little like uh link icon that's going to pull over um the references so you can go ahead check these out as you can see boom boom boom and uh yeah that's pretty much it guys this is L local search a fully local open Source uh perplexity

Alternative now if you're wondering okay which one should I use Josh you told me about this one you told me about the other four I'll leave links to those other four down below I've done multiple videos covering those as well as doing a review on them but I like I said at the start they all have their pros and cons one thing I'll say for a pro is like if you really just want to use local models um this one's mainly

Built for that so you could get some um cool stuff around that as well as it has its own uh rag system database Etc um now few cons anytime you're using local models but you know it's especially with this one cuz it has agents and whatnot I find or just in general but yeah it's definitely a lot slower like than something like perplexity of course all right if I was to use this or perplexity I mean I don't I don't think

This is as good as perplexity but I think it's it can I think it has its own strengths and some areas and some cool customizations it's something uh a cool tool to have in your toolkit or in your belt that you may want to use in certain situations and yeah if you want to see a more in-depth comparison between the other open source tools check out those videos down below but other than that guys that is it for this video if

You're new here we upload videos every single day on AI automation business growth Marketing sales open source tools Etc so if you like this type of content make sure to drop a like and smash that subscribe button to stay up to date with the daily uploads also too guys if you haven't already joined our free community strid community.com it's 100% free there's nothing for selling sale in there but you can just Network and connect with a lot of like-minded individuals as

Well as myself as well as I'm going to be posting free resources in those communities our free Facebook group free Discord channel so definitely check that out if you haven't already and then other than that guys I will see you in tomorrow's video let me know what your thoughts are in the comments down below if you triy the other open source tools which one's your favorite uh or do you just like perplexity other than that guys I will see you in

Tomorrow's video keep puzzling keep grinding and of course guys accelerate your stride take care