Build Your Own JARVIS with Jev (INSANE)

Henryk Brzozowski· 10 min· 1,885 words· 9 min read· English ·Watch on YouTube

This is the full transcript of Build Your Own JARVIS with Jev (INSANE), published on YouTube by Henryk Brzozowski. Every paragraph carries the moment it was spoken, so you can click any line to jump straight to that point in the video, search the whole thing for a word, or copy it out.

0:00Hey Jeff, can you open up Spotify for me, please? There you go, Spotify. Hey Jeff, can you play the song for me in Spotify? Playing. Hey Jeff, can you turn down the volume in Spotify and then open up [music] Slack for me, please? All done. Amazing. Uh hey Jeff, can you set a reminder to call Bob in 5 seconds,

0:27[music] please? Got it. I'll remind you. Hey, it's time to give Bob a call. Hey Jeff, uh can you now uh tell me who wrote Hamlet? William Shakespeare wrote Hamlet, one of his most famous tragedies from the early 1600s. Uh hey Jeff, can you now uh close down or lock down my computer, please?

0:53That's cool. Screen's locked.

1:00And there you go, it's locked. How insane is that? Now, I'm just going to get it to close stuff. Hey Jeff, can you close down my Slack and Spotify? Both sorted. That's so It is so bloody cool. This is one of the coolest and best things I think I have ever built. In this video, I'm going to break down exactly on how you could use it and potentially on how you could build it yourself. So, stick around and let's get to it. All right, so to make this work, we need at the

1:28bare minimum these three main things. Number one is we need Fish Audio, which is the sponsor of this video, and the voice that you heard was Hannah, and she's got that like super chill, realistic voice. And if you guys would have want to check out some of the other ones, Celine sounds really cool, Sarah sounds uh awesome, too. You know, if you are thinking of using something like this, like a like a your your own Jarvis, and you want like a male voice, you could certainly come in here and choose a male voice. You could even

1:53create your own voice it within here. So, you could clone your voice, you just have to talk to it for like 1 minute. I actually recorded myself in here before and I cloned my voice. So, if I for example say Henrik, I'm going to generate speech. Henrik. Henrik. FYI, if you're thinking like, "Yeah, I'm going to need to pay for one of those subscriptions." No, that is not true. Right now, official audio is currently for free. So, you can use it at least up until the end of November 2026 using this specific line, S2.1 Pro free. I think the only thing you have to pay for is this next thing, which is

2:27Jev. My name is Yeah. And it is a way to help us grab our commands or questions into a spot where something creates decisions for us. And these decisions get then turned into action for us on our computer. So, for example, before I said, "Hey, can you open up Spotify for me, please?" And it opened that up. Then it created a decision for itself. Does the user want to open up Spotify? Answer is yes. Then it would go and open that up. But, "Hey,

2:53can you open up Spotify for me, please?" Spotify, opening now. And um if you guys do want to learn more about it, I'm sure there's like a crap ton of videos out there right now on explaining on what Jev is. It is essentially our back end that controls the actions that we want it to take. What I would recommend getting is this thing called Open Writer. From here, you're going to be able to get an API key. And this API key lets us set up a fallback for things like when I asked it for example about Hamlet, when it can't take actions on my computer and it's outside of like its

3:25scope. What it's going to do is think, "Hey, I can't really answer this question." So, then it's going to go and ask an actual LLM. So, you guys can choose any LLM that you like. You can choose Haiku, you can choose Gemini models, you can choose your llamas or whatever you want. And you can get the LLM to create you the response. Now, I also, if you notice for the for the timer, it actually responded back to me very nicely. So, for example, if I say, "Hey Jev, can you uh tell me to book an appointment in uh 3 seconds,

3:55please?" Sure. I'll give you a shout. Hey, it's time to book that appointment. particular case, that was a timer that got sent through uh Jeff, then that got sent through uh I'm actually using a Haiku 4.5 just cuz I really like that model. Now, to some of the use cases that I think would be awesome for this. Um number one is what we were using just then, right? So,

4:22uh computer use / control. Uh later, I think this could be taken into like writing your own emails for you. I was just really excited, honestly, to get it to turn into an icon on my on my Mac. Uh have it open up in my dock. Um so, I'm sure you guys can think of uh more exciting things that you want to use it for your own computer use, for whatever you guys want to use that for. Another cool one would be like uh house control system. Instead of Siri or Alexa, you could be using this setup, which would be for free and running on your local

4:59setup, by the way. So, uh where when I'm running it from right here, it's actually not going the or the transcript is staying on my machine. So, we're actually using I'm just going to write here. We're using a small Whisper model, which only takes up maybe like 250 MB to run. Do an uh check every so often to see whether I've said, "Okay, Jeff." or "Hey, Jeff." And by the way, that's totally uh changeable. So, anyway, uh going back, house control system, uh you know, you could say instead of me like saying, "Hey, hey, Jeff, open up Spotify." you could say, "Hey, Jeff, can

5:33you turn on the living room lights?" Then, you could potentially use it, you know, as like a uh disability kind of helper to help you control your computer a little bit better. However, for me, what I actually would like to use a four, be to what do multitask. Hey Jeff, can you open up Spotify for me while I'm smoking? I don't have to like come to my computer and you don't press buttons. So multitasking, opening up stuff, and

5:56potentially lessen the screen time. If you guys can think of any other cool use cases, just put them down in the comments below. Now, we are going to get into the juicy stuff. So, if you are a person loves building this stuff, loves Claude code, Codex, stick around and I'm going to show you on more or less on how to put this together. So, number one step is you need an IDE. So, you could use Cursor, you could use any other IDE that you like. What I'm going to do is go down into the description

6:28and then grab this. Then, come back into Cursor, you're going to go clone repo, and then you're going to put that into any folder that you like. So, for example, I could put in my life get folder and I could select that as my repository destination. It's going to open that up for me. And now I have that project that I created for you guys. You can either follow the steps or you could just get your Claude code or Codex to set it up for you. So, for instance, you know, you would come into here, you would just type in Claude,

6:59and say yes. Then, you would grab this. You just copy this paragraph right here. You paste that in, press enter, and it's going to do all the magic for you. So, if you came here and you're like, "Hey Henrik, I'm using your Windows. Why the hell would I be watching you build it on your Mac?" Well, easy fix is, you know, you copy this and you say, uh "This guy was using uh this on his Mac and he everything so that it works here on my Windows desktop." And then just let it just let it rip. The things that you will need is your API key. So, I'm

7:33just going to show you real quick where they roughly exist. Uh you will then come into uh developer in Fish Audio. You will click on create API key. You click on create, copy that. Then go to typesafe.com, you apply if you don't have the if you don't have your API key yet. They exist right here though if you do. So you click on create API key, you create it, you copy that as

7:56well. Then you would come into open router, you'd come into something that looks like this. So you click on API keys on the left top right here. You would copy the API key. So once you got those three things, once you got your app working and you've you know you've discussed a little bit with the Codex and Claude, you come into this button right here that says keys. You press that and it opens up this new window you just paste all your respective keys into here. You press save. That gets saved to your Mac OS keychain. So like no LLM or whatever sees it so everything

8:28stays safe on your in your environment. So at the moment I believe Type Safe is actually for free. I haven't set up any payment methods on it. So I've been using a so I've used 1. 3 million tokens and it's cost me what's that? 4 cents? 5 cents? I was telling you guys S2 1.3 is now free for developers. So if you just I jump into here. I guess the only thing that I should really worry about is that when you create this app it's going to ask you for some permissions. So you might have to go and like click allow to use the microphone. Allow to open up

9:00Spotify for you. And then once you've got it working mine for example was put into a DIST folder as an icon like this. And then I just click on this and it just opens up the app for me. And then you can choose between hold option and then hey Jeff option. So thank you guys for watching. If you would like to sign up to Fish Audio I will I will give you absolutely nothing for it. Is there's a referral link down below. If you're like someone out there who just doesn't want to build this stuff for yourself then want someone like me to come in and just help you do

9:30it. You're more than welcome to also just book a call with me as also available right below. Anyway, thanks guys for watching and see you next one. Peace.

Where these words come from. This is the caption track YouTube holds for this video, written automatically by YouTube rather than by the creator. We read it, tidied the line breaks and laid it out so it can be read. The plain text version is at https://viewrankai.com/tools/youtube-transcript/7j8pcu3rq4k.txt.

All rights in this video belong to Henryk Brzozowski. Watch it on YouTube. If this is your video and you would rather this page did not exist, tell us and we will remove it.