🚀 Add to Chrome – It’s Free - YouTube Summarizer
Category: Tech Tutorial
Tags: AICloudDeploymentDockerRender
Entities: AI AcademyAnthropicAWSDaveDigital OceanDockern8nOpenAIRender
00:00
hello and welcome back to another video my name is Dave um now this video is being made today because of a couple of questions that come up in my school community's QA live session today um so we run an AI um Academy on school link
00:15
is in the description below and every couple of weeks we have a live Q&A session with me and questions came up today about hosting n10 and how you actually deploy your code and your AI agents online uh and
00:30
the cheapest and the best way to do that so in this video today I'm going to walk you through step by step how you'd go about doing that now there are several providers and actually one of the things I talked about today in the live session was using digital ocean which is a great way of doing it but actually in this
00:46
video today I'm going to be using render which is another Cloud platform very similar to digital ocean but there a few less steps involved so if you're a newbie to all this then this is a bit easier for you so what this is going to do is is it this will uh enable you to
01:02
spin up an instance of n8n um in a self-contained Docker image by a cloud provider on a cloud provider should I say um and the key thing is that means that this is then accessible from the public internet so if you actually want
01:18
to deploy these agents so that they are accessible online and can be used anywhere in the world uh particularly if you're selling this on you know as a as a an AI automation agency to a customer then this is a great way of doing it okay so my advice and as I said in my
01:35
live Q&A session today would be that you um have a cloud house cloud cloud hosting provider um with a Docker image with n8n and have them the separate accounts so they're the in terms of security they're um isolated and kept
01:51
completely separate so um if I have five customers at the moment in my agency each five of those customers would have different n8n automation agents and AI agents and those agents would all be installed
02:06
independently on their own clusters having no you know relationship to each other and um that that's just good for security and good it practice so anyway enough of the background let's get straight on the video so um what we're going to do is just move that window out
02:21
the way so we're going to head over to render.png this is the boring part of the video okay so once you are all signed up
02:39
to render service what we going to do is select a web services and then you're going to click an existing image now what we need to put in here is the URL to the docker image for n8n so we are going to find that now with Google
03:06
okay so what you're going to do is going to scroll down the docker page here and it's going to pick up this URL here that says docker.io sln n IO sln copy that come back to the render dashboard just paste this in
03:25
here right and click connect now we need to give it name now what I would suggest you do to keep this all nice and tidy and well well organized is to use the name of the customer who you are uh building this agent
03:45
for okay so you can choose your region here if you're in the US you've got lots of different options there you got some options also for Asia and for Europe as well now I'm based in Australia so I'm going to choose Singapore then we need to choose the size now if you're doing this just as a
04:02
hobby to mess around and to set this up to see how it all works then choose the free tier but just be aware that this instance can get spun down and then you have to manually spin up again which means that your agent isn't online so it may well work today but if it gets spun
04:18
down then tomorrow it won't be online and won't work you'll need to manually restart it which isn't good for production so obviously what we're going to do here in terms of production is just choose the $7 a month instance now the size of the image that you choose is
04:36
obviously fundamentally important to how complicated how many a API calls and how many agents you have running at the simultaneous time on that instance so um the good thing is obviously with um web
04:51
apps like render is it's easy to change the plan later on so we'll stick with starter to start with and we're going to scroll down okay and that's basically we're just going to choose on deploy web service so what you'll now see this this will take a few minutes and you'll see that basically this would install the N
05:07
Docker image on your new instance um on render.png
05:30
fully completed and then you'll see the URL that you'll be able to use okay so once that's all finished installing it says live there if I just scroll up a little bit further you'll see that my web service is listed as customer one and there is my custom URL
05:45
which I can copy and we can launch that in its own web browser and you see it's customer one and it's on reno.com and simply here all you're going to do now is register your account contact Vector Labs first
06:09
name click on next save those details just going to skip these quickly not interested for purposes of this video okay you can get the free license
06:26
key at the moment for the uh additional pay for features is going to skip that and there you go you basically have your own self-hosted instance of na10 which is available live online on the internet
06:42
you can now upload and deploy your workflows um you can so if you set this up correctly this will be done in the customer's name with their payment details so they're paying for this subscription um or if you have a different business model you can obviously have a set up in your name you
06:59
just need to be aware of where you could get caught out is if the agent is really getting activated lots and lots and lots and suddenly you're caught up in a tow spin having to increase the size of the ec2 instance um you if you're using AWS
07:17
back end um so yeah so that you got to be careful that you might end up with a situation where you're paying more money and you haven't got that in your contract that you can up the up the price depending on usage and also of course be careful of the cost of the API
07:33
calls as well if that's in your name you need to be very very careful of how many times the agent is being called and those costs for the API for open AI um can mount up very very quickly so personally I like to set this up with
07:50
you know my own API to start with and get the everything tested and demonstrated to the customer so they can see it all working and then what I'd like to do is once once it's deployed I have the customer set up their own API account open AI or or um anthropic and
08:07
depending which llm you're going to use and it's their account in their name they're the API key owner they pay the API costs and I just use the API key in their own workflow here so most of this then is in their name U and they are responsible for it you're just being
08:23
paid to set up the agent um and then you know maintain it if you have to uh make alterations and that kind of stuff but but basically there you go that that's how easy is and it's not expensive you know I've chosen a a service there that's $7 a month $7 usar a month and
08:41
that capacity of um compute Cloud instance would be enough to run a handful of Agents no sweat at all um and there you go that's it's as easy as that you obiously you can bookmark that URL now and give that to the customer and keep a record for yourself and that's it
08:59
that's how he easy is to deploy um AI agents using NN on the public internet so thanks very much for watching this video um I was going to do one on digital ocean as well um which I may do tomorrow but this basically covers you
09:14
know in essence what how you do it um so uh thank you very much for watching this video please make sure you like And subscribe and please feel free to check out my AI Community um which is called the AI Academy link is in the
09:29
description below got lots of courses on there and lots more videos and content and we've also got lots of free AI agent templates that you can basically just download and use in your own workflows and if you want to sell those on and make money from them then go for it no problem at all thanks very much for
09:46
watching this video and I'll see you on the next one bye-bye