The Best FREE n8n RAG AI Agents Course!🤖 Copy & Paste Templates Self Hosted Coolify, Ollama, Qdrant

What's going on everybody Josh poock here and in today's video I'm going to give you the exact stepbystep sop blueprint and guide to creating AI agents effectively and efficiently with n8n this is one of the most valuable videos I've ever uploaded on this channel and likely going to be one of the most valuable nadn AI agent videos you may ever see on the internet whether you're intermediate beginner or Advanced you're going to get some value from this we're going to go

Over AI agent fundamentals I'm going to show you how you can easily set up and self host nadn Lama postgress and cant on kify using my new done for you stride AI agents GitHub repo starter kit where I'm literally going to give you copy and paste templates and AI agent Frameworks that you can import into your n account right away as well as a complete 79 Pages in-depth document guide blueprint with all the screenshots links resources and Frameworks that you will

Need so you can get this going up and ready right now this may just be one of the most valuable n videos that you will come across on the internet let's dive right into [Music] it all right guys so as you can see this AI agent on telegram literally just booked an appointment for us automatically it retrieved some specific data from our Vector store and it did all this within 10 seconds quicker than an actual real life personal assistant could all

Right guys so everything that I got go over will be included Down Below in this free resource this document took me 7 days to actually get ready as well as get this whole training prepared for you guys so I really hope you enjoy it now this is called step-by-step SLP exact AI agent infrastructure that you can copy and paste to start building and selling AI agents for 2 to 10K using NN which involves no code I'm going to give you the

Done for you templates tools Frameworks examples and more all right within this document and training we'll go over different tools such as NN h postgress open AI pine cone and oama but there are additional tools that you can switch out and we'll dive into that in just a second so what will you get from this video now I know I said that this document is 79 pages and I'm not going to read through the entire thing especially the stuff at the

Start if you want to check this out like I said you can read through it thoroughly but I'm just going to go through skim through some of the most important points and then dive into that practical and applicable stuff for you guys so we can get this rolling right away now there will be time stamps down below and with that that being said we will go over all the foundations when it comes to creating AI agents first and this video may

Be a little bit long so I would encourage you to watch on 1.5 to two times speed if we're going over things that may be more practical or technical you can slow it down to normal speed but if we're going over things where you don't need to fully digest and you're just listening to it a little bit more passively maybe Theory or things you already know then you can speed it up so before we get started I just want to say

I've uploaded over 460 plus videos on this YouTube channel on different things revolving around AI marketing and business business and I truly believe that this is probably one of my most uh important videos that I've made I've personally spent over seven whole days documenting every single thing outlined in this document every screenshot line of code command line and Gathering all the most crucial info you need Etc so it's literally impossible for you not to get this set up as long as

You know how to copy and paste and I can almost guarantee you won't find any other step-by-step document for this on YouTube it's foolproof some people even may include this in some sort of course and charge money for it but you're going to get it 100% for free all I ask guys is if you do get some value from this video just drop a like let me know in the comments down below because that's what really keeps me motivated seeing that

You guys actually got value from this and then subscribe to stay up to date with the daily uploads if you're not already subscribed okay so what is the action plan that we're going to go over so we're going to start off with a quick background I'm going to dive into AI agent foundation and why n an is great for AI we're going to dive through prompt engineering and I'm going to give you the copy and paste prompt engineering Master template that

You can start using we're going to show you exactly how you can self-host the stride AI agents infrastructure on kify which is the n starter kit and then we're going to go over setting up the slack personal AI rag agent with copy and paste templates including postgress cuden olama and the telegram personal AI rag agent with tools which uses open AI plus pine cone and here's the screenshots of what these two workflows will look like in your nadn account I'm going

To walk you through exactly how they work and how to use them now who am I and why should you even listen while I'm the founder of executives.com we've worked with over 300 plus clients we've been in the marketing space for over 5 years now the author of client acquisition code a book on AI marketing business growth and the founder of stride agents.com fully autonomous AI appointment cers via voice email taxt and socials that's assassin service we're building around AI agents

And lastly just some proof that we do actually sell these services from anywhere from 2 to 10K plus now I will say if you are going to be providing this as a service to clients you're going to have to build upon this and iterate you won't just be able to use exactly what we provide to you here today this will give you a nice framework and starting point and you can improve upon it iterate and add your own value so why

AI agents AI agents represent the next Frontier in artificial intelligence offering autonomous goal oriented systems capable of perceiving their environment making decisions and taking actions to achieve specific outcomes here's why AI agents are crucial autonomy adaptability efficiency scalability Innovation and then AI agents are not just the future that the present the companies that adapt now will lead tomorrow all right and here's a link to the GitHub repo right here so I'll leave a link down below to this repo you're going

To want to St this I'll be making multiple videos around different AI agents that we're going to be adding into this repo so this is going to be 100% open source and if you guys want to add your own contributions iterations or whatnot we'll have some opportunity for that too down the line so our goal is to make this one of the most valuable AI agent repos so stay tuned for the future builds and trainings and then here's the link to

All the specific resources and templates for this training you're watching now so I'm not going to read through each and every point here I'm just going to read some of the main points on why AI agents are important so leverage and delegation at a fraction of the cost AI agents can work 24/7 without breaks increasing productivity data is everything so in the context of AI agents data is indeed everything the quality the quantity and the relevance of data directly impact the

Performance and capabilities of AI agents and prompt engineering prompt engineering is crucial for optimizing the performance of your AI agents it involves crafting effective inputs to guide the AI towards desired outputs and within this document you're also going to get access to the stride AI agent Master prompt engineering template all right so this template is very extensive and I want to stress here that you should not use this just word for word I would always highly suggest that you customize this

Template each agent and use case requires custom prompting to ensure you get the best results remember the quality of your output comes down to the quality of your input and your prompt and also keep in mind like this is a super super long prompt that pretty much has almost everything you would really want to think of in most cases you're going to want to shorten this to save on tokens if you don't know how to shorten it you could literally just

Give this to Claude or chat GPT explain to it what type of prompt you want and tell them to shorten it give it within a certain amount of words and you'll have a shorter in-depth prompt effective prompt engineering will mitigate biases and ensure consistent outputs okay so a few key components of an agents are tools AI agents rely on various tools to perform tasks and interact with with their environment these tools extend the capabilities of AI agents Beyond Simple Text generation

So within n you can access these tools right here and we'll go over these more in just a second and then Integrations are essential for AI agents to work seamlessly with existing systems and processes so these would be Integrations with things like your CRM with your email provider with your calendar all these different things are Integrations that you can give your AI agent access to now let's talk a little bit about the AI agent infrastructure so the architecture of AI agent

Systems is crucial for their effectiveness and scalability so to put it simply there's really three key areas there's department so AI agents can be designed to fit into specific departments such as customer service data analysis content creation marketing Etc and then workflows so workflows Define the sequence of actions an AI agent takes to complete a task they can involve multiple steps decision points and interactions with various tools and Integrations and then tasks so tasks are individual actions or operations that make

Up workflow the architecture should support the definition assignment and tracking of tasks all right so here's a diagram of the stride AI agent infrastructure right and here's a link to the diagram that you can access to and see it for yourself all right so Sops are very important standard operating procedures and they've always been crucial for business operations but in the emerging era of AI agents they are very very important so over time you want to start developing Sops standard operating

Procedures for every area of your business every process every workflow every Department Department should have its own sop here's some examples of different Sops we have in our business such as recruiting manager workflow salesprocess workflow because you have to think of it this way you can't actually delegate something to a AI agent if you don't actually have it very systematized and structured to begin with an example of an sop would be this exact document you're seeing also included here a few

Different other AI agent platforms just so you're up to speed and you're kind of aware where we're at in the space all right so we have crew AI autogen Lang Chang AI beam AI zapier Central there's a lot of different ones I even included a diagram here kind of giving a side-by-side view of some of the top ones here now why NN well n is better than make.com and zapier in terms of automations just in general it's open source so you

Can self-host it so it's literally free basically and it has really cool no code features so in terms of no code it is pretty much one of the best AI agent platforms you can build on it's been around for 5 years now and it doesn't have that steep of a learning curve all right so a few key links that you're going to need in this video are the GitHub repo of course we have N Docks we have self-hosting Dock and then

The N starter kit I had to Fork this n starter kit that n provides in order to make it integrate well with kify and be able to easily Plug and Play into kify all right next is the coolify docs right here and then free n AI workflow templates from the community are here and then the N form is right here all right so what's included in the N starter kit well a self-hosted n of course we have o llama so so

You can host your local models on olama and then cant and then postgress SQL you can use this for memory and your AI agent chats now there's two ways of using n you could use the Cloud solution right here but I would recommend using the self-hosted solution so the cheapest VPS provider is contabo so you can go ahead and sign up for contabo here if you want or you could use any other VPS provider of your choice now on contabo literally

The cheapest plan of $5.50 a month would do you could go on the higher plan if you want if you want a more powerful VPS now there's also a few different ways to self-host once you have a VPS you could self host and just install an directly on your VPS I recommend kify as it's the easiest possible way kify is a free open- source alternative to tools like versel or netlify and it's extremely easy to set up if I go to

Kif's website right here and I just click on self host as you can see there's literally just one bash command that you run right here in your VPS once you log into it in your terminal and that command is right here you run that and it's going to set up kify automatically for you on Port 8000 all right so once that's set up you can go to coolify and access it on Port 8000 so that will be your VPS and then

Port 8000 and then I would add a custom domain to coolify so you can see the docs right here on how to do that or check out some of my previous videos on coolify where I show you exactly how to self-host it and set it up integrate your domain and everything but it's fairly straightforward and fair simple all right so once you're in kify you're going to want to create a new project so click on projects then click on ADD right

Here okay and then you're going to want to enter your project name and click continue then you're going to want to click on production and then you're going to want to click on add new resource all right so let's do that right now so if I type in test right here and then click continue for my new project then I can click on production then I can click on add new resource all right now you have three options you can scroll

Down and either select NN or NN with postgress this is the direct one that coolify provides to you or you can use the stride and a end starter kit so either one works you could even try both and see what one works best for you this document and training will be complete and tailored for the stride AI starter kit so if you're not using it you will have to set up ol llama and the desired tools by yourself I've done a

Video on showing how to set up olama so I'll leave that linked down below if in case you want to use the coolify version so in our case we're using the starter kit so you're going to want to click on public repository and then you're going to want to paste this URL which is the GitHub repo into the public repository URL now you may even want to clone this repo or Fork it so you can make changes to the docker compose

As needed you may want a specific configuration that fits your needs so if you want to do that you can do so then you're going to want to click on check repository and then click on build pack and select Docker compose and then in Docker compose location paste in this right here and then click continue all right so I paste the URL in now I'm going to click on check repository now I'm clicking on build pack and I'm going to click

On on Docker compose now I'm copying this location path right here and I'm pasting it in right here and then you're going to click on continue then you're going to want to configure the project so you're going to want to point an a record of your NN subdomain to your VPS server IP address and then you could do the same thing for cant but you don't necessarily have to if you do this would expose it to the web so if you're

Doing this in production you'd probably may not want to do that or add some sort of authentification which you would probably have to Fork the um the repo if you want to do that okay so you would want to add your domain right here now you could hit generate domains like this and it would generate a dummy test domain but you're not going to want to use that in production and if you do do that then you're going to have to

Add a specific environment variable to allow an to be accessible on that dummy domain that isn't secure so you're going to want to click on the environment variables on the left hand side and then click on developer View and then you're going to want to copy and paste these environment variables right here okay so I'm going to copy these right here and this is the environment variable I was talking about if you are using the dummy domain then you're going to

Want to add na secure cookie equals false so if I click on environment variables then click on developer View and then you'll see these environment variables right here you're of course going to want to change this to a more secure encryption key as well as a JWT uh key right here you're going to want to change that you're also probably going to want to change your postgress DB password as well as your user right here and then this web hook URL

Is going to be the URL of your NN instance so put the exact domain that you used in the configure section for your NN right there in the web hook URL and then you're going to want to click on save all environment variables all right so once you save your environment variables you'll literally be able to just click on deploy and it will start deploying the N instance all right so here's an example of an N starter kit that I deployed

On kifi that is currently running and this is what we're going to be using for the remainder of this video all right so if you're completely new to n I've outlined here the different AI agent structures that you can use so there's chat models right here and here's all the different chat models there's memory right here and here's the different memory options and then here's Vector stores and here's the different Vector store options now like I said we're only going to

Be using specific ones in today's video we're going to be using cant and pine cone as well as postr for memory if you want to use Zep I have a video on how to use Zep right here so I have some technical stuff here about how Vector stores work in case you want to read up on that so indexing querying filtering Etc and then some applications of vector stores such as semantic search recommendation systems image and audio retrieval and anomaly detection

And then here's a diagram so data then vector embeddings and Vector data and then Vector database so just so you know Vector stores allow an AI model to reference relevant pieces of documents useful for question answering and document search Vector stores are specialized databases designed to effectively store manage and query High dimensional Vector data there are crucial component in many Ai and machine learning applications particularly those involving similarity search and recommendation systems and they also go hand inand with embeddings so

Embeddings are dense Vector representations of data that capture semantic meaning in a way that machines can understand and process they are fundamental to many modern Ai and machine learning techniques and you don't need to know all the little technical stuff especially at the start regarding vectors and embeddings I don't want to confuse you guys I do have some more technical information and descriptions of That of vectors and embeddings down here in case you do want to look into them as well

As where vectors and embeddings are based off of I've have included the research papers here such as word to VC Bert sentence B Etc all right to put things into simple terms vectors are lists of numbers that represent things mathematically and embeddings are special vectors that represent complex objects like words or images as points in a high-dimensional space capturing their key features and relationship so it's essentially a way for AI to to do retrieval augmented generation which is Rag and when

You query the rag system it's going to pull the relevant data from your database using vectors and embeddings so just quickly if we look at a rag system on a high level there's really two phases of it the one phase is data insertion so upserting or inserting data into a vector database so let's say we upload data into a Google Drive and then we insert that into a vector database right here we're using an embedding model right here to embed the

Data into uh the vector database right here and then when we query the data so we ask our rag agent a question we're embeding the user's question or our question SL query and then the rag system uses different mathematical equations to basically compare the similarity between our query embedding and the similarity of the actual insertion embedding of our documents to be able to tell where it should actually pull data from all right so you can imagine if you have tons and

Tons of company data or whatever type of data you have it's a lot easier keeping that in a rag system opposed from having a huge context window and then you're going to reach the context limit regardless but you would go through a bunch of tokens you're going to have to provide that every single time so regag essentially tries to solve that it's not perfect yet but it's getting there it's getting better and better and this is the effective way to use

It and to deploy these systems into your business and making them applicable when working with larger data all right so there's actually three phases total of your rag AI agent there's first which is data preparation phase so this involves things like data cleaning and conversion chunking embedding indexing and storage next is the retrieval phase so this is the vector similarity search relevant information retrieval and then the third is the generation phase so context augmentation llm processing response generation El presentation and

You're going to see this in action in just a second I just want to give you guys a high Lev overview okay all right so once you sit up Up N with kify you can go to n and the default Port of n is 5678 just in case you need that but like I said you can either set up your domain within coolify like I showed you or you can generate a test one if you're just testing things out and then

When you go to production you can connect your domain okay so once you're in there in the home section you can either go to credentials or you can go to the workflows and I'm going to provide you and connect your connections through there all right so you're going to want to add a credential so once you go to credentials you'll be able to search up any single credential you need to so first one we're going to talk about is AMA all

You're going to need to do is change the base URL from Local Host 11434 to ol 11434 and then click save or connect and then save that and I have the instructions right here now if you're not familiar with AMA it's open source LM models and you can access any open- Source llm model from olama within your VPS so let's say we wanted llama 3.2 how you would actually install that onto your setup would be you're going to want an SSH

Into your server once you're in your server you're going to run the command Docker PS okay just copy and paste this it's going to list all your Docker containers that are running you're going to see your o llama do Docker container and you're going to want to take that name or ID and you're going to run this command exactly now you're going to want to change this to whatever the model is that you're pulling you can go to AMA and you'll

See the exact Command right here for whatever model you want so we can see here it's ol llama run llama 3.2 and we would just change out this area right here where it says container name or ID and you would put the container name or ID just put one of the other and you're just running Docker exec dasit and then your container ID and then olama pull and then llama 3o whatever right so this Docker exact allows you to run commands

From that specific container Now by default within the stride n starter kit we've included the NCH embedding model from oama as well as llama 3.1 so you don't have to install those but if you do want additional ones you can follow this guide I also have a video showing you how to do this without the AI starter kit so okay but this is basically showing you just if you already have Ann and you're watching this video and you don't want to

Use n starter kit because you already have an n uh scenario set up that's perfectly fine you could actually follow this uh tutorial right here now for the postgress connection you're going to do the same thing and add a new connection for postgress So within postgress you're going to put the host to postgress the database is going to be the same database name that you use in your n environment variables so this could be anything but this depends on what your

Environment variables are the ones that we set up earlier the same thing for the user and the same thing for the password once you do that the port's going to stay the same 5432 and you're going to click connect and then save so you can use any API key for cant and then the URL is going to be cant and then 6333 for the port and then just click save all right so it's going to look like this for the cant

URL the API key can literally be anything all right to connect your Google services accounts I left the links to the docs right here they're very straightforward and they have the exact screenshots documentation they even have a video here for this if you have any questions and you want me to add additional screenshots and documentation in this document let me know in the comments down below if you're having any issues connecting your Google services but I think it's pretty good and

Straightforward with the doc so you're going to want to connect your Google calendar account as well as your Google Drive you'll notice with the workflows that I'm providing you guys today we use slack and telegram reason being is because I know most businesses either use slack and many people use telegram so you may be using one or the other you could use both or you could switch this out for something like WhatsApp something like Discord but I wanted to cover the

Two main ones here today now another thing I want to touch on before we get into these workflow builds is we're also using the Lang chain node for one of them um the reason being is because at the time of this video some of n's Vector store nodes don't support upserts because you can't get defined IDs to go through embeddings which means you'll have duplicate Vector embeddings if you try to run the same content through a second time so we use

The Lang chain node and this is something I don't see too many people talk about it's it's essentially a lang Chang node where you can add an embedding node as well as add some code in there so I'll show you that in just a second and that's what we use for currant to be able to upsert this and this code did take me a bit of time to figure it out and troubleshoot it to get it to work effectively and properly

So I'm going to be giving this to you away for free of course within the workflow build and I'll show you that in just a second so without Ado we're going to get into the actual stride AI agent starter templates so the first one being the slack personal AI agent assistant now this one's going to be using the local models and currant so basically everything we just set up is going to be put into use so to get started with the

Stride AI agent templates you're going to go to the GitHub repo here and then you're going to click on the o1 folder and then you're going to click on n8n SLB backup right here and then you're going to click on the workflows folder and then you're going to see four different Json files you'll actually see five you can ignore this random one you're going to see the slack personal AI agent the telegram personal AI agent and then the two tools which

We have the create calendar events tool as well as the get calendar events tool so if I click on the slack personal AI agent right now you'll see we have this Json code right here all you're going to do is copy the raw file like so and then you're going to come over to NN and you're just going to paste it in that's going to paste in the actual n template right here you can go ahead and just rename it and

Then save it and you're going to want to create separate workflows for each Json file and paste those in so in total you should have four different Json files two different agents and two different tools are you tired of pouring thousands of dollars into appointment Setters only to watch leads slip away imagine having a team of elite sales agents booking qualified appointments for you around the clock no more wasted time on training no more frustration with performance and no more draining

Your budget on inconsistent and expensive call centers introducing stride agents AI powered appointment sets that work 24/7 never get tired and book appointments while you sleep trained on thousands of successful conversations our AI agents El perform human teams at just one tenth of the cost join the ranks of businesses that doubled their appointments and booking rates in just a matter of weeks don't get left behind in the AI Revolution visit stride agents.com now and transform your entire sales process with Cutting

Edge AI technology it's time to accelerate your stride with AI agents now first things first I included the slack connection documentation right here now I will say it is a little bit more complex than some of the other connections okay so this one I did specifically outline everything and I will walk you through it right now so first things first is you're going to need to create a slack app so you can choose your existing slack workspace or create a new

One and then you're going to want to go to api. slack.com slapps right here so this will show all your apps you can click on create a new app and then click on create from scratch you can enter the name here and then select your workspace and create the app so next is setting up your Bot so under the settings section click on on basic information and then fill in the display name and default username at the bottom and save your

Changes so if I scroll to the bottom here we can see display information you can change the name of your Bot next is we're going to want to add the permissions to our bot so in the leftand side menu we're going to click on o off and permissions then we're going to scroll all the way down to Scopes and we're going to add the required bot Scopes okay for this example you'll need at least the app mentions read the chat write

And the Channel's history so you can see these right here so if you scroll down the Scopes you'll see the bot tokens right here you'll see that I added a bunch just because you can add as many as you want but you're going to need at least those three if you want to add additional ones you can if you're encountering any issues with Scopes or permissions just you could add them all for example once you do that you're going to install

The bot to your workspace so if you scroll all the way back up and then click on install app so click install app to workspace and then authorize the app so this is what it's going to look like install to your workspace name see mine says reinstall to workspace but yours will say install and remember this location because we're going to come back here shortly all right the next step is to retrieve the bot token so after you install it to

Your workspace go back to the O off permissions page and copy the bot user oo access token it should start with xxb Dash and this is the access token for your slack connection so when you're creating your slack NN connection here you're going to paste in that access token right here and then click save to connect all right the next step is the events subscription so you're going to want to click on event subscriptions on the left side menu and scroll

Down to subscribe to bot events and then click on ADD bot user events and then add the app mention event and then click save changes so you'll see event subscriptions then you're going to want to toggle the events on right here and then you're going to want to add the bot subscription event right here and make sure you have app mentions here you'll see I also added these other two you could add these as well you could add as many as

You want but you're going to minimum at least have to have app mention okay you'll also notice that there is another step right here for the request URL this is going to be your web hook URL from n now where do we get that well in this slack agent if we click on the trigger right here you'll see right here it says web hook URL now I'm not going to click on this because it would show the web hook URL um

But once you do there will be a test web hook URL as well as a production I would recommend using the test to start out and then you can switch to the production once you validated this whole process now for this specific step you're going to need the web hook URL set up in your environment variables within your nadn so if you didn't already set up that uh web hook URL if you didn't already connect your domain to your VPS you're

Going to need to do that you can't be using the generated coolify test domain right now you're going to need your actual domain so if you need to go do that go do that and then redeploy the kify NN instance and then make sure that you have your web hook URL showing that is uh the one that you set in your environment variable okay once you do that you're going to copy that test URL and you're going to paste it in

The request URL right here and then click on connect now once you click on connect before you click on connect actually you're going to need to go to your na scenario and you're going to need to click on test step this is because slack is going to automatically send a request to your n URL and it's going to need to pick that up so if you're using the test URL make sure you click on test step then it's going to connect

And then you're all good to go so from that point on you're going to need to go back to the install app location and reinstall the app to your workspace and then authorize it after you're done that you need to go to the channel that you want your Bot in for me you can see that I made a specific Channel called AI agent and you're going to want to type in for SL invite and then at your bot name that's going

To invite your Bot to that channel so to quickly go over where you should be at now you should have your slack connection set up you should have your olama your postgress and your cudr all set up and connected now you should have your slack rag AI agent template loaded into your account now I'm going to go over this template with you so you can know exactly how it works and where you can customize it if needed so to start off

We have that slack trigger so if I want to go ahead and clicked on a run and then sent a message to my slack agent it would pick this up we'll do that in just a second here I also have this chat trigger right here too just in case for testing purposes if you want to test it out using the chat here next you'll see we it's going to an AI agent it's a tool agent right here so if I click

On the tool agent like I said we're using the tool agent there are different types of Agents here I'm not going to go over all these right now but we're using the tool one now you'll see that there is a prompt right here so there's the text right here this is a prompt and you'll see it's basic it's simple just saying you're a conversational agent you can go ahead and read it but it's basically telling it what tools it has able

To use and a few different other things um you can go ahead and change this or optimize it if you'd like so we're basically telling this agent that if we ask it a question that requires additional info then it can check the vector store and you'll see the vector store prompt in just a second and then if the the user asks to either check the availability of a calendar or or book a calendar event then it's going to use these two

Different tools and we we'll go over those in just a second so the chat model it's using is our o llama chat and this is what we set up and you'll see we're using llama 3.1 the latest version next is the postgress memory chat so this is basically going to allow it to reference the uh other chats that you've been using you can change the context window length to be more but right now it's uh set to five and then you'll

See the table name this is just what's going to add to your post postgress table and then the specific key the session key so you can either Define that below or take from previous node okay then we just have two basic tools the calculator tool and the Wikipedia tool this is just in case you ever ask it a question where it needs to do some basic math or search something up on Wikipedia there's also additional tools like I mentioned that you

Could add and improve upon this agent such as HTTP requests to make it able to send different API requests okay next we have our Vector store so if I click on Vector store you'll see that the prompt is returned documents related to AI questions rag prompting Etc you can limit to how much it's actually going to pull for results right now we just have it set to four but you could change this and the name that we're using for this Vector

Store is called documents all right so next we have two tools the get calendar events tool and the book calendar events tools right so the G calendar events is calling upon the G calendar events workflow and there's a short description here explaining when to actually use this tool and you can go ahead and read this or change it around if need be all right the source is database and then the workflow we're pulling it from the list which is the G

Calendar events right here the workflow that you would have imported and then the field to return is response and I'll show you that in just a second and then the schema type we're using generate from Json example and we're providing an example right here for start time and end time and this is what that workflow looks like so we have an execute workflow trigger right here so once this is being called upon it will run this trigger and then we have

A a Google Calendar event so this is why you need to have your Google calendar connected so we open this up we're using our connection the resource is event the operation is get many we're selecting our calendar right here and then the limit to return I have it set to 50 and the options right here is the before and the after so we're pulling the Json end time and the Json start time so that's why we specified those uh Json events

Start time and end time example in the other workflow all right then we have an aggregate node right here so we're actually putting uh output in fields to data right here and we're including all fields and then we're returning the field right here in with the value in uh with the name response okay so this is what we're returning for this workflow so that's where it's looking out for this response field right here okay so then it's going to take that

Response and feed it back to the AI agent all right we have a similar thing set up right here for the book calendar event so say if you ask the AI agent to book a calendar event hey book a calendar event with Josh here's his email at this specific time so here's the description you can go ahead read this maybe change it around a bit if need be we're doing the same thing for the source here workflow is the create calendar

Event workflow the field to return is response again and then the schema type is a generated Json example right here for start time end time attendees and then about which is just a short description okay so that create calendar event workflow looks very similar executed workflow trigger and then we're creating an calendar event this time so you can see here we're using the connection right here the resource as event the operation as create and then the calendar we're selecting our calendar

And then we're putting in the start time right here the end time and and then some additional Fields is we have all day for no and then attendees is the Json attendees email that we're pulling from the uh AI agent all right the description is right here and then the meeting location so right now we just have it set to Google meet but you could go ahead and change this and then summary right here is the about and then send all

Updates to all the different guests here you can go ahead and change this if need be then we're doing the same thing we're aggregating those results and then setting the field right here for the response all the data is going in the name response and that's going to be sent back to our AI agent and it's going to be able to confirm that it booked the meeting all right so those are all the tools and embeddings and memory and chat that

The agent has access to once it's done with the query it will send a confirmation message right here back to slack now on the other side of things we have the data upsert into Vector database so this is using the olama embeddings and the currant Vector DB so there's two ways that we're actually doing this one is by creating a file in Google Drive so if we create a new file into a specific folder destination so this is why you're going

To need your drive located or connected in here so you'll see the poll times is set to every minute and then the trigger is changes involving a specific folder and then you're going to select your folder right here and it's going to be watch for file created so this is actually when a file is created you'll see this one is the same thing except this is for changes involving a specific folder once a file is updated so when a file is

Created or updated it's going to trigger this then this is going to set the node for the file ID right here so this is pulling in the file ID that you either added or created and then here is a download file node for Google Drive so we're actually downloading the file for the resource and the operation is download of course using the file ID right here to find that specific file ID and then for the options we have Google file conversion

Set for Google docs to be transferred into text document and then we have an extract from text to files so we're extracting the text from this binary data field that we just downloaded and we're extracting that into text and then here we have a langang node which is our currant upsert all right so this is using our olama model this is the same olama model we're using up here we're using the noch embed model I'm not going to go over all

This code with you right here but we essentially have some code right here that will um basically upsert this cudr document into your database like I mentioned before currently at the time of recording this currant doesn't have a native upsert function within the N node so we have to use a lang chain node with custom code to do this and we have some specific inputs specified right here okay so let's go ahead and test this out so I'm going to upload

A file right here to our Google drive folder all right so I just uploaded an agent Q document from another YouTube video we made now if I go to this created file node and I click on test you'll see that we got the agent Q file right here so next I'm going to run this set file node and then I'm going to run the download file node then I'm going to run the extract text node and then I'm going to run

The cant Lang Chang upsert node okay so the upster is done if we click on the node right here we'll see our data from the previous node and we'll also see right here our schema we have the uh previews of the different chunks right here so if this all works let's check out our cant dashboard and see if the collection is there so like I said before you can access the cant dashboard at Port 6333 so it will either be your

IP address or if you set up a custom domain you can go there and we refresh and we can see here documents for the collection and we have 67 points six segments one sh shards and if we go ahead click on this and here we can see all the different points from all from the document that we just upsert it okay and like I mentioned we have 67 points and to prove to you that the upsert actually works and it's not

Just inserting I'm going to go ahead and rerun this cant upset search and we should still only have 607 7 points if we have double then that means it inserted it and there's a duplicate if we have 67 then the upsert upsert worked correctly all right so the upsert is done let's go ahead and check our cant database here and after refreshing we still only have 67 points so that means our upsert node worked all right so now that we got

The upsert into our cant database we're going to actually prompt our rag AI agent on slack and we're going to test out all the specific tools so I'm going to say please tell me everything you know about agent Q the AI agent tool and book a call with Josh joshp poock gmail.com for tomorrow at 100 p.m and make sure you at your agent to begin with all right so you can see the node picked up our slack message right here all

Right so the slack trigger picked up our message and now I'm running the AI agent as you can see it checked postgress chat history and now it's using the olama model right here now I will say the old llama models locom models take a lot longer that's why I included the telegram version as well which is using open Ai and and pine cones so if you want a faster agent that you're using on the day-to-day I would recommend using the open

AI model okay so we can see here it's checking the vector store tool right here olama cant all right so the git calendar events and book calendar events tools just ran now we're still finishing off the vector store right now all right so as you can see here the whole entire rag system ran we got the post press memory AMA the vector store and then the slack message with the calendar booking events and you can see here that it booked the

Calendar event right here for the meeting with Josh and here's a screenshot from my calendar as you can see call with Josh for tomorrow now I will be 100% honest this whole entire system took a while and it's not perfect especially when you ask it multi-threaded questions like you saw what I did there it didn't get it perfect by any means it didn't it did struggle with P the information about agent Q it did book The call but it isn't perfect

Like I said and that is one of the main reasons is because we're using the local model um so you'll see in just a second when we use open AI it's going to be a lot easier a lot quicker now to connect to telegram here are the docs right here as well as the bot API docs for telegram but I actually outlined exactly everything you need to do to create a telegram bot it's very simple you just enter new bot uh

Once you start a chat with bota you can literally follow these instructions it's super super easy it's way easier than the SL back and once you create your Bot it's going to give you this access token API right here and this is the API that you're going to use for your telegram connection so you're going to paste that in here and then just click on Save and then for your pine cone connection I left the docks here the API reference um

A few different things with pine cone so it has indexes metadata and Nam spaces Nam spaces are pretty cool kind of allows you to categorize um your different collections into your index similar to kind of like what collections are in uh cudr and here's the na and connection docs for pine cone you can create a free pine cone account at pinec cone. once you create your account you can come to API Keys over here create a new API key and that's

What you're going to use to connect to pine cone in NN all right it's very simple in Pine Cone once you're in Pine Cone you're going to create a database then you're going to create a index by clicking on create index you're going to name your index input the dimensions and Metric so depending on what embedding model you're using you can click on the setup model right here and it's going to show you the different embedding models and then you can

Click on the one that you're using for this example we're using text embedding three small from open AI so it's going to put the same dimensions for that embedding model and the metric which is cosine and 1536 then you're just going to click on create index and there you go you got your pine cone set up you're just going to connect it to NN and you're good to go for that so now you should have your telegram setup and connected your

Pine cone setup connected all right so I just inserted a document into our pine cone database here I'm going to click test workflow right here and I'm going to send that same message into telegram for our bot now I'm going to run our agent and look how much quicker that was this was literally a fraction of the second of the olama model all right so I was having some issues with the rag system it was basically just because I was using

The wrong model here than the one that we were using for the insertion so I fixed that so it should be good for you guys as you can see here I asked about agent q and it did pull successfully this is information directly that we inserted into our database and then just to double check make sure it worked correctly I asked it a multi-step um question right here so in a short paragraph uh tell me about agent Q the AI agent

Tool and book a call with Josh for this Tuesday at 6 p.m. so it's summarized agent Q right here this is exactly from the data that we inserted and then additionally we have a call with Josh for Tuesday at 6:00 p.m. so you can be obviously um change around these prompts make it a lot more optimized and specific for your specific use case but these tools should give you a really good starting point for building out AI agents all right it's

Going to introduce you to the concept of using different tools we only made two different tools here but I would definitely recommend creating multiple tools this should really be a starting point that hopefully give you some different ideas like creating different tools for checking email sending email checking CRM you can really create multiple tools and have this agent basically be like an actual personal assistant all right so you have the basic framework here go ahead start creating start building and guys

Like I said this is just the start I'm going to be building upon this GitHub repo so make sure to start it and stay tuned for the future builds we're going to be doing more NN builds as well as different actual custom coded bills such as crew AI maybe open AI swarm you know autogen multiple different builds build surrounding AI agents I'm going to be giving you guys this all for free so I really hope you guys got some value from

This video like I said that 79 page document with additional uh information regarding na AI agents n is in the description down below I also included some additional AI agent templates that you can check out from the community right here so you can see all these different AI agent templates some around web scraping some around using super base all these different ones here there's probably about eight or so I also included additional videos and resources that you can check out some

Made by me and then some additional reading for Na and surrounding agents so you can improve upon uh you know really your knowledge base around this specific topic guys you can really start selling these type of agents and Frameworks and you know systems and infrastructure to many different types of businesses or using them in your own business whatever you want these agents are huge right now and they're going to continue growing I'm actually serious guys when I told you that I've

Been ending this video for the last 7 days so if you did get any value from this video if you let me know in the comments down below that's going to bring a smile to my face you know that really motivates me when you guys get some value from these things and then also too guys I'd really appreciate if you like the video and uh you know hit that subscribe button if you're not already subscribed already also too guys if you

Haven't already joined our free community stri community.com I'll leave a link down below our free Discord Channel and free Facebook group you can check that out I'm really trying to build a community here guys with like-minded individuals and the AI space marketing business sales whatever and that's one of the main reasons why we're also starting this open- Source GitHub repo where we're going to be providing extreme amounts of value guys so definitely check that out and also too guys if you

Run a business whether you're a marketing agency business owner B2B B Toc whatever the case is you run a call center and you need AI agents in your business that's one thing that we specialize in when it comes to AI appointment sets AI voice callers AI texter we've been implementing it into other business is and they've been seeing extreme amounts of success this is like a night and day thing you can actually replace real call center Reps for a fraction of

The cost like I'm talking on10th of the cost and believe it or not they're actually lots of times more effective because they're not lazy they're always positive and they're trained on thousands of successful conversational Frameworks so check out stride agents.com I'll leave a link down below and I'll also leave a link to my calendar to book a call in case you're interested so we can see if this would be something that would make sense to implement into your business or not

We built our own software so if you are an agency yourself and you want to resell our service or wh label it that's something that we can also talk about or you can completely white label our software it integrates within go high level and everything like that so book a call down below if that's something of interest to you other than that guys I will see you in tomorrow's video if you're new to the channel we upload videos every single day

On topics like this thank you guys so much for the recent 8K subscribers road to 10K on the way and guys if you do find any bugs or anything of issue within the N work CL or within the setup let me know and I'll make sure to update it for you guys other than that I will see you in tomorrow's video guys keep hustling keep grinding and of course guys accelerate your stride take care