Build Your Own AI-Powered Brain: Step-by-Step Guide with OpenWebUI

Summary

Highlights

• Dave shows how to build a personal AI using OpenWebUI to replace corporate digital brains like ChatGPT. 🤖
• Enjoy complete ownership and control over your data by using AI models directly on your computer. 🏠
• Flexibility to switch and try out different AI models like Llama 70B hosted via Grock. 🤟
• Access premium models from OpenAI using API keys and unlock more functionalities. 🔑
• Step-by-step installation using Docker and Olama makes AI setup user-friendly. 🚀
• Mobile access is achieved with Enro, letting you take your digital brain wherever you go. 📲
• Web search capabilities are integrated, allowing the AI to provide updated information. 🌐

Key Takeaways

• Unlock the power of a personal, privacy-focused AI with OpenWebUI. 🔓
• Maintain control over your data by running AI models locally on your computer. 🖥️
• Explore a variety of open-source AI models without being tied to a single provider. 🌐
• Learn to set up Docker and Olama software for an easy AI setup process. 💻
• Access a powerful AI assistant from your mobile device and enjoy AI on the go. 📱
• Avoid corporate-owned digital brains and opt for a customized AI solution. 🚫🤖
• Use advanced open-source models for free with Grock's hosting service. 🎉

Overview

Chapters

• 00:00 - 06:30: Introduction to Building a Digital Brain with OpenWebUI In the introduction chapter to 'Building a Digital Brain with OpenWebUI,' Dave Wearing, from Digital Brain Base, introduces the concept of creating a personal digital assistant using OpenWebUI. He expresses his enthusiasm about replacing corporate-owned digital brains, like ChatGPT, with a personally controlled system. Dave promises a step-by-step guide on setting up one's own digital brain and displays his own creation, which he mentions is visually similar to ChatGPT.
• 06:30 - 10:00: Owning and Controlling Your Digital Brain Locally This chapter discusses the concept of owning and controlling your digital data locally without relying on big tech companies. It highlights the benefits of maintaining privacy by keeping personal data on your own computer, ensuring that no internet connection is required and your information remains private. The chapter emphasizes the empowerment of users to manage their digital information autonomously.
• 10:00 - 13:30: Downloading and Using Different AI Models The chapter discusses the advantages of using AI models locally on one's own device rather than relying on large tech companies over the internet. It highlights that users can select and switch between different AI models seamlessly, allowing for more personalized and controlled computing experiences. The emphasis is on privacy and customization by using local installations of AI models.
• 13:30 - 18:00: Setting Up Software: Docker and Olama The chapter discusses the flexibility of working with open-source models, highlighting the ability to download, own, and control any available model from the internet. It addresses various trade-offs, such as speed and answer quality, illustrated by an example where the user switches to a different model to improve response time and answer satisfaction, despite it being slower due to its larger size.
• 18:00 - 25:00: Installing OpenWebUI and Initial Setup This chapter discusses the initial installation and setup process of OpenWebUI. The speaker explains the option to run models locally on their computer without an internet connection, and mentions the necessity of reconnecting to the internet to access more powerful or proprietary models. These models become available through connection to a host provider, which is apparent when refreshing the interface.
• 25:00 - 35:00: Downloading and Installing AI Models with Olama This chapter discusses the process of downloading and installing AI models using a service called Grock, which hosts open-source models on their servers. The focus is on the Llama 70B model, a large model with 70 billion parameters, in contrast to smaller models with two or three billion parameters. The narrator explains that downloading such a large model onto a personal computer would result in slow performance, if it works at all. However, by connecting to Grock via the internet, the model can be accessed and utilized with much greater speed.
• 35:00 - 45:00: Using Grock for Access to Large Models and APIs The chapter titled 'Using Grock for Access to Large Models and APIs' discusses how Grock enables the use of various large models, including OpenAI models, efficiently and with ease of access. The speaker elaborates on the speed and convenience of using such models via APIs, which allows them to ask questions swiftly and maintain control over the AI they choose to interact with. This setup empowers the user to integrate any intelligence they wish into their digital tools while retaining ownership and control over the AI's usage and application.
• 45:00 - 53:00: Adding Web Search and Mobile Access to Your Digital Brain The chapter discusses adding web search and mobile access to a digital brain setup. The speaker demonstrates how they have integrated various technologies, including utilizing OpenAI's GPT-4 model, to enhance their digital systems, all while maintaining control over their data. Importantly, they've managed to do this without any cost or obligations. To illustrate the process, the speaker switches to a computer without pre-installed software to provide a step-by-step setup guide.
• 53:00 - 54:30: Conclusion and Invitation to Join the Community The chapter provides a guide for joining a community by setting up the necessary software on one's computer. It emphasizes that the required software is completely free with no hidden costs and requires minimal technical knowledge. The first essential software mentioned is Docker, accessible through docker.io. The process is designed to be user-friendly, ensuring anyone can follow the steps without technical expertise.

Build Your Own AI-Powered Brain: Step-by-Step Guide with OpenWebUI Transcription

• 00:00 - 00:30 hi guys Dave wearing here with digitalb brainbase decom and today I'm pleased to show you the digital brain that I built for myself using open web UI I'm also going to walk you through how I set it up step by step so you can stop using corporate owned digital brains like chat GPT and start using your own digital brain instead let's take a look okay so here we go you can see my digital brain here that I built using open web UI and as you can see it looks very similar to chat GPT
• 00:30 - 01:00 uh but with one huge difference which is that I own and control everything that you see here on the screen uh and if I want to run this 100% private I don't even need an internet connection to do it I can keep everything all my personal data and sensitive information here on my computer instead of sending it to a big tech company okay so let's go ahead and ask a question here please please give me a meal idea for my kids that is is simple easy and
• 01:00 - 01:30 healthy as you can see I'm still not connected to the internet here and here it is giving me that meal idea that's coming 100% from my computer and the things that I've installed here instead of sending my question out on the internet to a big tech company another thing that's super cool is I'm not limited to any any one model I've
• 01:30 - 02:00 downloaded several other models and I can really download and own and control any open- Source model that's available on the internet so for example if I didn't like the answer that gimmo was giving me for this particular query or was too slow Etc there's different tradeoffs depending on the model I could ask 53 so here we go with that model you can see that I changed it right here and this model is a little slower because it's bigger
• 02:00 - 02:30 but there you go that's also coming from my computer without any connection to the internet um but if I do want to use faster models or I want to use some of the closed Source models like uh like those from open AI I can do that um I have to turn back on my internet connection in this case connect back to the internet refresh the page here and now you see in my drop down I have tons of different models because I've hooked up to a host provider called
• 02:30 - 03:00 grock that hosts these open source models um on their serers so I can access them with much uh greater speed so llama 70b is a very very large model it's a 70 billion parameter model versus the two and three billion parameter models that I was using before and if I downloaded this model onto my computer it would run very slow if at all but here because I've got it hooked up to grock through the internet you can see how fast um that is that's really even
• 03:00 - 03:30 faster than chaty BT and I can even use if I really like the open AI models there they are I've got those hooked up uh with the API so I can ask open AI same question and there you go fast and uh easy so it really gives me the control that I want to use any model that I want any intelligence that I want really I should say with my digital brain that artificial intelligence uh from anywhere while still maning maintaining a great degree of ownership
• 03:30 - 04:00 and control over the data that I'm and if everything I've shown you so far isn't awesome enough uh I can even I've even done everything that I've showed you here except the gp4 model from open AI uh I've hooked all that up and I'm using it 100% free with no strings attached so let's look at how I set it up okay guys so I've switched over to my wife's computer which doesn't have any of the software on it so I can show you exactly how to set this up step by step
• 04:00 - 04:30 uh you just need a couple pieces of software both of which are completely 100% free no strings attached and to run one command on your computer um and this is these are steps that anyone can follow regardless of technical experience so the first software that we need is called Docker if you go to docker.io other types of software without having
• 04:30 - 05:00 to do any coding or setup it just makes it really easy and fast to get up and running so once that's downloaded you'll find that in your download files and you just click on it to install it and run it just like you would any other application that you're running on a Mac same similar process if you're running on Windows just whatever the standard process is uh whenever you download any software okay and then once that's downloaded we want to go to our applications find it
• 05:00 - 05:30 and go ahead and start it up yes we'll open that up so just go ahead and use the recommended settings here and we can skip all this we don't need to do any of that and we've got
• 05:30 - 06:00 Docker up and running on our computer right here now you can see next step is we're installing a software called olama olama is an open- Source piece of software that allows you to run all the different AI models that I showed you previously in this video so just download it detects again what type of computer I'm on it's got windows and Linux there if you're on another type of computer I'm just going to go to our downloads again and open this up and do the exact same process that you would
• 06:00 - 06:30 when you're downloading uh any other types of software you can see now it's been downloaded so we can click on it yes move to Applications so it's in our applications folder and just click on the install here and just click finish there okay and now once we've done that we see that
• 06:30 - 07:00 we have both olama and Docker up and running on our computer and we can go to the last step uh and that last step is getting open web UI installed on the computer uh and this is the most technical part but stay with me because uh it doesn't matter your level of technical expertise anyone can follow this so we're going to go to docs. openweb ui.com SL getting started I'll include a link to this in the uh description section wherever you're watching this video and
• 07:00 - 07:30 we're going to copy this uh snippet right here by just clicking the copy button and then we're going to open up the terminal application on our computer whether that's a Windows or a Mac it's called terminal on both and there it is and we are just going to paste that right here and click enter and that is going to to
• 07:30 - 08:00 install open web UI on our computer and once that's done we just have to sign in and We're Off to the Races so we'll give that a minute to finish up and we'll be right back okay and now that that's downloaded we're going to open up Google Chrome and go to Local Host 3000 and you might say Dave I thought you know we didn't have to be connected to the internet to do this and you don't I'm actually not connected to the internet right now and if I try go to a
• 08:00 - 08:30 website you'll see no internet so this is accessing the local resources through the uh software that we downloaded onto our computer and even when we sign up for this new account uh it is going to save all of this on our local computer it's not going out to the Internet so sign up here and create an account and now I've logged in and I get this initial screen with just some of the things that
• 08:30 - 09:00 they've added to the software recently I click okay let's go and here I am but there is one last step which is I need to add the intelligence right now I don't have any AI models downloaded onto my computer so I need to download the models that I want to use along with this digital brain that I've just set up using open web UI okay and this is where olama comes in really handy so we're going to have to
• 09:00 - 09:30 turn our internet back on to download the models onto our computer and we're going to go to ama.com and here's the models tab that they have that shows you I mean I think there's hundreds of models here there's just page after page of them uh and you can click into each of the AI models to learn about them here's llama 3.1 which is an open- Source model family from meta and you can read about that the
• 09:30 - 10:00 sizes all the evaluations Etc um but for the purposes of this video just to get you started I recommend starting with the Gemma or Gemma I actually don't even know how to pronounce that uh Gemma 2 uh model the two billion parameter model because it'll run on pretty much any computer some of these bigger models uh they're either going to run very very slow on the average computer or not at all um so this is a good one just to get um started with and start playing around
• 10:00 - 10:30 with it and you can see if you scroll down the page here you can see this Command right here o Lama run Gemma 22b to get this two billion parameter model that I was showing you at the beginning of this video and you go right here and you just paste that in and you pull it and it starts downloading it from the olama backend that we downloaded so that's where this olama software comes in it gives you access to
• 10:30 - 11:00 download uh hundreds of different uh open-source AI models to add the intelligence component of our digital brains okay now that that's downloaded we can start using that model along with our digital brain but before we do let's go ahead and turn off the internet so we can prove that this is now 100% under my ownership and control on my computer with no internet connection needed so give me a short poem that I can give to my
• 11:00 - 11:30 wife and there you go we're now up and running with our own digital brain complete with the ability to download and control whatever open-source AI models for the intelligence portion of our digital brain that we'd like now depending on the power of your computer you may be able to run larger models locally without an internet connection but uh you know it can get pretty expensive the Machinery that you need to
• 11:30 - 12:00 run some of these larger models is not something that the traditional um consumer laptops currently are very good at uh so if you do want to try out some of the more powerful models um then I recommend checking out grock it's G Q with a Q not a k the K one is uh Elon musk's AI through Twitter or x uh this is different from that and what grock does is they host all the main big open- Source models on on their servers and
• 12:00 - 12:30 allow you to access uh their inference uh which comes at a super fast speed and when you're accessing as an individual um it's basically free um I haven't they have a free tier and I use it all the time and I haven't run up against any of the rate limits that have required me to pay so seems like at least for now you can use this uh along with their digital brain completely for free okay so now let's look at how to set this up um for some reason I was n able to get the grock API to work without doing the open
• 12:30 - 13:00 AI API first uh in the open web UI interface so uh but that's also free to set up so if you haven't got an account already go ahead over to open.com and set up an account it's free you just click sign up there I already have an account so I'm going to sign in and then once you're signed in you want to search API
• 13:00 - 13:30 and go to the introduction page here and go down to API Keys you can also just go to once you've signed in platform. open.com for/ API Keys uh and we're going to create a new secret key here give it a name I'll call it open web UI create and then just click create secret key
• 13:30 - 14:00 key it's going to ask you to verify if you're human and now you've got your secret key so once you have that don't share that with anyone I'm going going to delete this one when I'm done you don't want anybody to be able to use uh your API and then we're going to go back over to open web UI we're going to click on our name here in the lower left hand corner we're we're going to click admin
• 14:00 - 14:30 panel and then settings and then we're going to go down to connections and we're going to paste that open AI uh key that we just generated yours will be blank here if you haven't set this up yet you paste that right there then you click save here at the bottom and if that's working when we click the chat button here and then go to select model we should see the GPT
• 14:30 - 15:00 models which is all uh open AI models here uh and if we um disconnect from the internet and we refresh we see that those go away and the only model that's showing is the model that we downloaded on our computer and that we were using before and now we're ready to go ahead back over to grock and set up the gro API okay so if you don't already have a grock account just sign up it's free click sign in right here and then click
• 15:00 - 15:30 Start building and the first thing that we're going to do is go to documentation and we're going to scroll down to where it says requesting your first chat completion we're going to want to make sure that curl is selected here and then we're going to copy this URL right here but only up to the point where it says V1 slash we're going to leave off this chat completions part we're going to copy that and then we're going to go
• 15:30 - 16:00 back over to the admin panel in our open web UI interface we're going to click this plus sign right here beside the open web beside the open AI key that we just set up and we're going to paste this link that we just copied from grock here where it says API base URL now once we've done that we're going to go back over to grock we're going to click API keys create new API key give
• 16:00 - 16:30 it a name don't share this with anyone I'm going to delete this one once I'm done here and go back to once we've copied that and click done go back to the admin panel and paste that API key right here click save here at the bottom right hand corner and now when we go to create a new chat in open web UI and we click
• 16:30 - 17:00 select model we'll see a bunch of other models besides the GPT models here including all the big open source models from meta which are the state-of-the-art uh competitors to Clos Source models like open AI GPT so there we go now we can chat with let's use this llama 370b model give me a meal idea for my kids that's simple and and
• 17:00 - 17:30 healthy bang and there you go that's coming from grock and you saw how fast that was um and if we can if we again disconnect from the internet and refresh the page here go to new chat and we'll see that only the model that we downloaded on the computer is available now and we're completely we can run this still completely offline tell me a joke see what it comes up with
• 17:30 - 18:00 [Music] here why don't scientists trust atoms because they make up everything okay so we're all set up with our own personal digital brain that runs locally on our laptop that we own in control using open web UI and we've really just scratched the surface here there's so much you can do with open web UI to make your digital brain more powerful um you can customize your own model you can upload different models
• 18:00 - 18:30 like train them different ways you can have your own uh prompt templates and use prompt templates from other people that use the open web UI software and that they've created uh in their Community there's tools you can build on top of this and use tools that other people uh have uh have built as well through the community uh functions are a way that you can execute uh programs like if you've heard of agent capabilities those types of things I mean this software can
• 18:30 - 19:00 really uh do it all and it's getting better and better uh all the time but there's two more quick things that I want to show you before we sign off here um to really help you make full use of your digital brain and those are uh adding the ability to search the web uh with your digital brain and adding the ability to access this locally running digital brain this open web UI workspace that runs on your laptop over the Internet securely from
• 19:00 - 19:30 your mobile phone so you can have everything that we've set up here running on your computer at home and then when you're on the go you can take your digital brain with you uh through a secure connection connecting to your local machine at home and again this is for free and can be done regardless of technical ability so let's take a look okay so to set up our digital brain
• 19:30 - 20:00 built using open web UI to search the internet we're going to Google Google programmable search engine and go to this site right here and click get started if you don't have a Google account already you'll be prompted to sign up for one uh and if you once you sign in you're going to see this page right here we're going to click add we're going to name the search
• 20:00 - 20:30 engine and then we're going to select search the entire web we're going to turn on image search we're going to click I'm not a robot and then click create and the first thing we're going to do is we're going to take this series of numbers and letters right here behind where it says CX equals and we're going to copy that
• 20:30 - 21:00 and then we're going to come back over to our open web UI and we're going to click on our name and go to admin panel settings and then we're going to click on web search we're going to enable this if it's not enabled already and then we're going to drop down select Google psse which stands for programmable search engine and we're going to paste that
• 21:00 - 21:30 uh search engine ID series of letters and numbers that we just copied right there and then we're going to go back to our programmable search engine we're going to click customize we're going to scroll all the way down to the bottom where it says programmatic access we're going to click get started and then we're going to generate our API key and again don't um share this with anyone and I'm going to delete this one when I'm done with this video select whatever my first project is fine here
• 21:30 - 22:00 click next and then click show key and copy that and click done and then go back to your open web UI and paste that API key right here and then click save now we should be all set here and if we go back to open web UI and select SE a model and we say
• 22:00 - 22:30 please give me the latest polls for the US presidential race which is happening right now as I'm recording this video it will say I'm unable to provide real time information but if we click this plus sign and click web search and then ask the exact same question now it's searching the
• 22:30 - 23:00 web and it's not only going to give us uh what it believes the answer to this question is but it's also going to give us the sources that we can click on to go and see more about where it's gotten that information from so now we're all set up to search the web and we can move on to the final step which is to access all of this via our mobile phones okay so to do that we're going to go to enr.com and inro allows us to set up a gateway
• 23:00 - 23:30 to our local machine uh that we can access securely over the Internet so we can use our open wey driven digital brain on our mobile phones so sign up for a free account okay I'm already signed up so it's going to take me to the connection page I'm going to click download here it's going to detect the type of computer that I have I'm going to click download here save that to my downloads and once that's downloaded I'm going to
• 23:30 - 24:00 copy this Command right here I'm going to go back to my terminal and paste it there it's going to ask me for my computer password and once I enter that password it's going to ask me if it's okay to give terminal access to the downloads folder and I'll click allow there and now it has extracted that file and I'm going to run
• 24:00 - 24:30 this command here copying this go back to the terminal run this command and it says the authentication token has been saved to the configuration file so I'm all done here now I need to scroll down and right here where it says under deployer app online now I'm going to copy this and go back to my terminal I'm going to paste that there but I'm going to change this to from 8080 to Local Host
• 24:30 - 25:00 3000 okay now I have inro up and running on my computer I can go here to the endpoints page and now this is the link that I can now use to access my open wey digital brain running on my laptop through a secure connection via my mobile phone so let's do that now
• 25:00 - 25:30 I'm going to copy this link and take it email it to myself so I can open it on my mobile phone okay so here I am on my iPhone and here's the link that I emailed to myself the enro link and if I click on this you'll see that it brings up my open web UI interface to sign into and I click sign in there and I've now logged in and you can see
• 25:30 - 26:00 that the chats I was doing before are here on my mobile device because I'm accessing my open web UI that's running on my laptop so here's the same chats that we were that I was just showing you on my desktop and now if we select a model till me a joke there we go why do skele since fight each other they don't have the
• 26:00 - 26:30 guts these are definitely dad joke types okay I hope you're as excited as I am to now have your own digital brain so that we can stop using corporate owned digital brains like chat gbt and start using our own digital brains instead I'll be creating more videos like this in the future so if you're interested in uh learning more about the more advanced features of open web UI as well as seeing how I'm using my digital brain uh like And subscribe on the platform where you're watching this video and if you're
• 26:30 - 27:00 interested in taking really deep Dives and following along the complete process as I go about building my own digital brain I've created a free community where I'm sharing all of the things that you've seen here and more and I'd love it if you join me at digitalb brainbase tocom and I've included a link to that site in the description section hope to see you there soon and thanks for watching