n8n AI Agent Chatbot Assistant Basic Overview 🤖 OpenAI, Anthropic, Vectors, Embeddings

What's going on everybody in today's video we're going to be looking at how you can use open AI or really any llm anthropic Etc in your n8n scenarios all right how you can actually set these up what are the different functionality within the different nodes NN has a really really cool way of doing things that helps you if you want to do things like rags and bunch of other cool stuff we're going to dive right into it [Music] all right so

We're going to walk through some of the basics of the n n AI um nodes and how it all works and um yeah I'm going to be doing more in-depth videos showing you specifics of you know setting you know specific things up and I'll show you that in just a second what I'm talking about um but I know when I first started using n and I looked at the AI Bots and you know the different nodes I was kind of like

Oh this is cool but there's also I mean if you're used to something like make.com or if you're used to I mean just anything other than um maybe do like actually building out Rags or using uh AI within like the command line or within um your own SAS products or whatever you may have had some trouble understanding how um to actually use it with an n8n show you that in just a second but for those of you who don't know what

8en is it's essentially like an alternative to maybe zapier or make.com bit more I guess Advanced if you want to say um a little bit more of a learning curve it's not too difficult and um it does have provide you a lot more freedom though I really really love this tool um there's not enough videos on YouTube about this tool showing you different things so hopefully um you can check out some of my other videos too if you're new to it

And you'll get some value there setting it up um I've covered videos on how to actually self-host it within your own VPS Etc cuz the the great thing about n8n is one yes the functionality is super super good you have a lot more freedom than other apps even make and zapier but also too you can it's open source and you can self-host it so you can put on your own VPS and where other tools you know zapier may charge you a

Couple thousand to year whatever for depending on how much scenarios you're using it could be even more if you're using a lot of scenar or using a lot of executions or whatever um same thing with make even though it's a lot cheaper than zapier it price trust me can stack up could be spending a few thousand even 5,000 $10,000 a year if you're using like a good amount of EX scenarios right um like for example I spent uh just over like

200 bucks on or just around 200 I believe uh on make.com scenarios in the the following month right so obviously it can stack up um good thing about NN is you know if you're self-hosting it you get unlimited executions unlimited runs so it's awesome anyways check out some of my other videos if you're new to NAD but once you're actually here we'll get into like if I go to um we'll just show some tests um here so I can just do

A trigger for the workflow and you'll see that M or nadn has this Advanced AI section and you know it has other sections too like you can run code do all these other cool things and um that's all awesome but we're going to be looking at the advanced Ai and building autonomous agents summarize or search documents Etc so this is really cool one you can look at their templates here they provide templates part of their community so you can check these

Out they have some cool templates and we even look at some of these just to give you a sense of what can actually be done here so we can filter there's different categories here like yes you can look at engineering or devops or marketing or business or Finance or SEC Ops whatever we're going to obviously look at AI we'll look at these in just a second and get some good ideas of use cases but before we even get there I want

To just kind of go over some of the functionality we have so here we have ai agent and we'll actually cover that actually we don't need a trigger as a as a click cuz this automatically gives us a trigger as a chat and you can trigger it with anything it could be web hook you don't need this chat here but this is basically like if you wanted to um you could talk to it uh but let's just see here so there's

The AI agent there's also I go here there's open AI where you can do like create an assist uh where you can do like the standard stuff like um you can message an assistant you can message Chat so it's just a basic node there um basic llm chat you could use this for certain things it's just basically this except basic it doesn't have memory or tool and then question and answer chain and then Advanced summarize chain okay and then we also

Have some other AI nodes so we have document loaders language models this is just all the language models so we have anthropic WS Bedrock AZ Azure uh Google Gemini Palm Gro hugging face mistro o Lama uh open Ai and we have memory so this is actually a lot of these are so we can see like Memories over here but um we have miscellaneous Vector stores embeddings tools Tech Splitters retrievers output parsers and Etc and I'm not going to be able to

Cover like every single little thing in this video I just want to give you a good understanding so you kind of know what these are what it is and maybe some what you can actually do and then in future videos I'm going to show you in depth maybe more about setting some of this up setting up uh different um you know uh Vector stores or embeddings and all that good stuff um you can use tools like super base um pine cone

Um so I'm going to be doing more videos in the future on super BAS as well and um and then yeah so over here you can choose like for AI agent you can choose the model so let's just say I chose open AI okay you would connect your open AI account you could just do that with the API key same process goes for any model then you could select memory so standard one is the that's the easiest just this window buffer

Memory there's also these other options so motohead radus chat memory Zeta and then zap um so like this one's the easiest um at least that's you know that's what they say um and window buffer memory so you would take the session ID you could take it from previous node automatically or you could Define it below you would get the key from the previous like say if we were running something here um we could get our own session ID key here and

This key is used to to store the session ID in the memory so essentially this is just like memory so it will me remember previous chats um okay and there's so this is good um if you're if you use something like make.com and I did a video on this a few days ago um you may be used to like if say if I'm messaging an agent on open AI I can it can give me the thread ID and then I can

Just like say if this was a no this is just a node you got the thread ID from the messaging the agent then in the next node I could just reference that thread ID and I wouldn't have to re-input all the like the a longer context window of what they previously said so I found that really useful because you can save tokens now in n8n um you can't put the thread ID you can't put the thread ID so I did a

Video a few days ago where I showed you how to actually get that same process done just by using the API docs and using HTTP requests to open AI then you have to send the thread ID um so you can do it that way but um you know that's that's how I kind of did it for some but this using these the actual n um the nodes so I mean you could use these ones so retriever so contextual compressor retriever um

And to be like I haven't even experimented with all these yet so I'm just going to kind of tell you guys what I've learned so far and I'm going to go more in depth in future videos but I was kind of confused when I first like started seeing all this because I was honestly used to something like make.com where it's very simple you just have the AI module a node Etc and then you it's very simple naden it's it's still very

Simple once you get the hang of it um but it's you know it's a bit I guess a bit more of a learning curve just a few extra things like memory here I haven't tried out all of these different memory stores like some like the thing is this one's obviously free that's why it's easiest window buffer memory these ones like this one is open source but um I don't think they're there there's no they're not uh this is like depreciated or

Whatever they're not uh currently updating in and support or something like this I saw on their GitHub so don't know if I would necessarily maybe use that one I don't know um there's reddish chat memory Zeta and zap so those they're all they have some pre-plans but then there's also uh paid options but yeah so those are a couple memory options tools so this is like let's say you can use like calculator um so if you want to do calculations custom

Code tool so you can write Json or pyth or JavaScript or python Ser AI to access Google search data Wikipedia Wolfram Alpha and then you can also connect custom nadn workflow tools so I could do this and I could actually um connect it to one of my other na scenarios and you put the workflow ID here ID here and then it this agent would actually go ahead run that scenario too if needed um but yeah like if we were to run

It and you would put Wikipedia here and it actually answered uh it was a question that I asked it about something that it would get from Wikipedia then we would when we would go here and see the output we would actually be able to see um that it actually did do use this tool all right so this is like basically giving it different options doesn't always necessarily like if you put a calculator here and ask it like a question not regarding

Math then it won't need to use the calculator and same thing with like the workflow or the what like any of these tools right they're just things that you can give it all right and let me see I guess you can give it multiple tools so that's cool as well and put like Wikipedia you can give it a calculator if you want memory so you could just use window buffer memory and here's where you select the context window length so a

Number of previous messages to consider for the context and um yeah so let's see there's also AI um other AI nodes so open AI chat memory Lang Chang code so you can connect it to Lang Chang I'll be doing more videos on Lang Chang in the future and parser so parsing the output um and there's I also want to show you guys so document loaders default data loaders uh where are we oops I've been really looking into super super base lately

So I'm excited to get more into that and to be able to show you guys some videos on that in the future so stay tuned for that but if we go over to the templat section here I just want to show you guys some different things that you could do so let's just say I'm going to filter this by popularity all right so here we have an AI chat agent so this is actually you can see who created some of these

So this is scrape uh web pages and summarize with AI um telegram chat bot so here we'll look at take a look at some of these right here so scrape and summarize web pages with AI so we can click to explore here and basically scrape latest pul gram essay so we would do an HTTP request get request to that website then we are extracting essay names all right Source data Json property okay so extraction values and the CSS selector so doing

That then we're splitting the items not going to go through the whole technical stuff here if you want to use this you can easily just import this on into your make account you can see here then once we get the essay text we're using a summarized chain right here with open Ai and AD Json input loader okay so I guess this is probably Json text yeah and then recursive character splitter and extract title and then we're merging the title and summary

And then we're cleaning up the data let's take a look at this one this is a little bit more um a little bit more basic and and easy probably to you know look at for a first time and understand everything so this is just a AI agent chat so conversational agent model chat jbt memory window buffer we're using the chat history series session key from the previous um open AI module and then context window length five and then a tool is

Our Ser API account and output parser I guess you could add that too um but yeah I just kind of wanted to go over some brief you know things regarding the n8n uh you know AI features the nodes kind of what they do in in a very broad sense we're going to be diving more in depth um in future videos all right where I actually show you exactly certain things I'm building out how I'm actually using these tools how I'm using

Um you know building out the databases in the back end whether it's on super base or whatever um to actually like have um info to train the bot on and and what not and how I'm actually doing that so I'm excited for those videos um and I hope this gives you somewhat of like an understanding of what the functionalities within NN are if you're new to an or if you saw the AI features in N you didn't really know what they

Actually do um yeah this is the memory is just to summarize the model obviously selects the model memory is like basically you can either just use this window buffer which is keeps it within the workflow and it's essentially remembering whatever chat you have here but if you use like a different memory type like one of these you can like put uh get it from the uh the database whether it's on Zada rdus Etc and um and then trying to find super

Bas yeah and then Vector Stores um you can use tools like super base and some other ones for Vector stores we're going to get into that in future videos all right so like I said just wanted to give you guys overview of this great tool n an and the AI capabilities of it you can use Lang Chang you can use open AI you you can use uh mistol you can use Brock you can use um quad anthropic so it's really awesome

If you actually are able to learn these things um the possibilities are really endless and if you are do get stuck while learning these things like if you have a question about what something does my suggestion yes you can ask me in the comments down below and I'll definitely um be more than happy to assist you but um always ask chat GPT or always ask Claude you know that's what they're there for be your your guide your tutor and yeah you

Can learn things super super quick with these tools it's really really insane so other than that guys um if you're new here and you haven't already joined our stride Community it's a free group of like-minded entrepreneurs individuals uh AI enthusiasts Etc link for that will be in the description down below we have about 750 to 800 members in our stride Community also too guys if you haven't already checked out my book client acquisition code go check that out link will for

That will be in the in the description down below and then other than that guys if you're new here make sure to smash that subscribe button we upload videos every day on AI automation um you know growth growth hacking marketing Etc um so stay tuned for that I will see you in tomorrow's video guys keep hustling keep grinding and of course guys accelerate your stride take care