DEA: AI Coding Assistant with Potential

Deva's DEA is a fast-evolving AI coding assistant with potential to build applications quickly, despite some current bugs.

### DEA: A Promising AI Coding Assistant with Rapid Development and Local Model Support
* 🤖 DEA is a fully-featured AI coding assistant that is an **open-source clone of Devin**, with a user interface that includes a **chat, browser, terminal, logs, and settings**.
* 🚀 The project is still in its early stages but is progressing quickly, with **continuous improvements in the interface, support for more models and search engines, and overall development speed**.
* 💡 DEA allows users to code applications using **zero-shot prompts**, and it currently uses **Vim for coding**.
* 📦 Users can select from various models such as **OpenAI GPT-4, GPT-3.5, Mistral**, and others, and run DEA locally with **AMA**.
* 🛠 The installation process involves **cloning the DEA repository**, **creating a Python environment**, **installing dependencies**, **setting up the front end and back end**, and **configuring environment variables**.
* 🌐 Users can select search engines and models, and run DEA with different APIs such as **Grok**, **Mistral**, and **O'Lama**.
* 🐛 While DEA is still somewhat buggy, especially when using certain models or browsing the web, local models work perfectly, and the project shows **great potential for the future of AI coding assistants**.

there it is. Look at that, unbelievable. Very, very easy. So that's at least one of Devon's main demos recreated in DEA. All right, I finally got DEA working, and we now have another opsource clone of Devon. This is a fully-featured AI coding assistant that comes with everything you need to code with AI. So this is the interface. On the left, we have the chat. On the right, we have the browser and the terminal. We have our logs right here so we could see everything, and we have settings, which doesn't work. So just like open Devon, this is a very early project, but they are moving incredibly fast. Just a few days ago, I couldn't get it working at all, and the interface was completely broken. But now the interface looks a lot better. They support a lot more models, a lot more search engines, and things are really coming along. So I'm super excited to see where this project goes. So today, I'm going to show you a little bit about it, and then I'm going to show you how to install it and use it.

so this is the calculator app that I just built. It is not pretty. These numbers are not in order, and it's horizontal, which is kind of weird, but it works. This is all done in zero shot. A simple prompt was all I need. Build a calculator app with a UI. So defa says I'm outlining a plan to build a calculator app with a UI. Here's the plan: choose the programming language, step two: set up the environment, design the UI, ucss, etc., etc., all the way down to I think I can proceed without searching the web, which is great. And I have completed my task. And here we go. They are actually using Vim to do the coding. So you can simply use this to build any application that you want. And again, it is quite buggy still, but it is making great progress. Up in the top, you could see whether you have access to the internet or not. I think that's whether you actually have the server up and running. You have the token usage. You could select your search engine. They now support more search engines. I'm using.

Google, I'll show you how to set that up. You can also select your different models so they support the Claud models Gemini Gro, which is super fast. Although I couldn't get these to work well, and there was actually a race condition issue, so I suspect Grock might actually maybe be too fast for DEA, but we'll see. They have the mistol models, which is great. We can power it locally, of course, with AMA, and we have open AI GPT 4 and GPT 3.5. I'll show you how to set all of this up, and then after it was done, here it is. Here's the index.HTML, here's the script.js, and the style.css, all done in under 1 minute. Now, granted, this is a very simple application, but it shows that this is working and that this project has great potential. So just to show you that it works, here's 8 * 8 = 64. All right, let me show you how to install it. Now, the first thing you're going to do is open up VSS code, and we are going

to CD to the desktop or wherever you like to store your new projects. From there, we're going to clone DEA. So, git clone and then the Deva GitHub URL. And if you switch back to the Deva GitHub page, you click this green code button and then copy right here. That's where you can find that URL. And I'll drop this link in the description below. Then hit enter. Now we're going to use cond to create a new Python environment. So, condac create DN Deva python equal 3.10. Hit enter. Now, I already have one named that, so I'm going to go ahead and overwrite it or remove it. And you won't see this error unless you already have that environment. So, go ahead and click okay. Okay. Now that that's done, you're going to highlight this right here, and it's going to be paste cond activate Deva. Hit enter. And then you can see it's active because it says so right there. Now we're going to CD into DEA, and then we're going to type which

python. So, we're just making sure that we're always installing things with the right python environment. So, which python? You're going to copy whatever gets output there, paste it, then type `pip install -r requirements.txt`, and then hit enter. All right, once that's done, we're going to do `playwright install`. `D-wi DS` and to install playwright, you're going to type `npm init playwright@latest`, and then hit enter. And then it says, do you want to use typescript or JavaScript? We'll use typescript. And if you don't have npm installed, you can probably use Brew to install npm, but simply just Google it and you'll find out it's very easy to install npm, and then hit enter a few times, and then it will install playwright. All right, done. Now we're going to install with playwright. So, `playwright install -D with-deps`, hit enter. Then we're going to `CD` into `UI`, and then we're going to `npm install`. Next, we're going to do `bun runev`.

and again, if you don't have bun installed, you got to install it. Just a quick note on how to install it: you're going to type curl DFS Capital SL https colon bun. Shinall pipe and then bash and then hit enter. But I already have it installed, so I'm going to skip that. Then we're going to type Bun Run Dev to spin it up. Okay, so that worked. There's our Local Host 3000, but we're not done. We actually need to spin up the back end. So go ahead, click right here, open up a new terminal. We're going to conda activate DEA again. So then we're going to CD into the DEA folder. So CD desktop Deva, hit enter, and then we're going to spin up the back end by typing Python dea. piy and then hit enter. And sometimes this takes a few minutes because it's downloading a sentence Transformer model. Okay, so loading sentence Transformer Bert models loaded, and it should be almost done. Olama not available, okay, because we're not running an AMA model right now. So I'm going to show

you two more things. Once I show you that it's working, I'm going to show you how to run it with Grok Mixol and just how fast that is. And then I'm going to show you how to power it with Olama. All right, so Deva is up and running. Switch back to that Dev tab. Right over here, we're going to hold down our Command and then click the Local Host button. And there's DEA. All right, so it's up and running, but we're not done. We actually need to fill out our environment variable. So come back to VS Code. You're going to click the Explore button, and we're going to click Open Folder Desktop, and then just open DEA. Now, unfortunately, I think that actually closes all the previous terminals, which is kind of silly, but it does. So let's make sure that we are using our Conda environment DEA. All right, so this time, for some reason, I have a config dot L file already, and we also have the sample. Last time I did this, it wasn't there, so maybe I did something.

different this time, but look for this. Just rename the sample. config.msi. So here's all the different models that you can use, and then here's olama. If you wanted to power it locally, for now let's try grock. So I'm going to put my grock API key right there. I'm going to click save, and the next thing you want to do is grab your search key. So we are going to use Google search, and go to developers.google.com custom-search slv1 introduction. Then right here it says get a key, click that button. We're going to select a project. I already have one Deva, click next, and then we're going to show the key. We're going to copy the key. So we come back to VSS code, we're going to highlight this, and we're going to paste it right in there. Now we also need the search engine ID. So you're going to come to programmable search engine. goole.com control panel all, and you can create a new search engine right here.

once you create it, you're going to click into it, and you're going to grab this search engine ID. So copy that. You're going to come back to VSS code. You're going to highlight this and paste it right there where it says Google search engine ID. Then you save, and we should be more or less done now. So let's spin up our servers again. So we're going to do python dea. py hit enter, and it should spin up faster this time. And while that's going, we're going to click this plus button. We're going to cond to activate DEA again. We're going to CD into the UI. And then we're going to do Bun Run, and that'll spin up the front end. Okay, so Dev is up and running. DEA is up and running. Going to hold down command click on this link, and it should be ready to go. There we go, sweet. So select project. We're going to click new project. We're going to type game of life 'cause we'll create that one again for the

select search engine. Go ahead and click Google for select model. I'm going to select Grok Mixol. And then let's try to create it. Create The Game of Life in Python with a UI. Then hit send. Okay, so this should be lightning fast. Yep, browsing the web. I don't think I've actually browsed the web before, so let's see if this works. Okay, so I got the same error as I got last time. Playwright Sync API inside the sync IO Loop. This seems to only happen when I'm using Grok, so that's why I'm kind of thinking this might be a race condition. Okay, so I couldn't get Game of Life working with Grok Mixol, and it was actually due to a bug in DEA, but I was able to get it working with ChatGPT. So let me show you that now. So I'm going to select a new project. I'm going to call it Game of Life 'cause that was one of the demos of Devin. So I'm still going to use Google, and I'm going to use GPT-4 Turbo. So I'm

going to keep it pretty short. Create a Python project that makes the Game of Life, and we can see all of the output right here. So, it's probably waiting for OpenAI to reply right now, and I'll also show you Mixol on Grock just to show you how fast that is. All right, so it is working. I'm thrilled to outline a plan. For some reason, it's outputting twice right here in the chat, but that's okay. So, research The Game of Life, set up the environment, etc., etc. I think I can proceed without searching the web. That's good, and DEA has kind of a personality. So, okay, diving into the Game of Life. It's a classic. First up, I'll refresh my understanding of the rules. All right, so it is Echo waiting just to make sure the terminal worked. All right, and it says it completed the task. That was really fast, and it created a requirements.txt file. So, let's test it out. Let's see if it works. So, I am opening a new VS Code. I'm in the Deva.

folder. Then I go to data projects, Game of Life, and then I'm going to hit open. Here are the requirements, and here is the code. So let's see if it works. So I'm going to open up the terminal, and I'm going to do pip install dasr requirements.txt. Looks like most of it's already installed. And then python main.py. There it is, look at that, unbelievable, very, very easy. So that's at least one of Devon's main demos recreated in DEA. Now, let me show you how to set it up completely locally. Go to ama.com, click download and install olama if you don't already have it running. Now, I'm going to leave the front end up, but I'm going to shut down the backend server. I'm going to click plus, and I'm going to type olama list, and that should give me a list of the olama servers running if it is up and running. So there we go, we have mistl latest, that's fine, and that will also spin up ama if you don't already have it running.

okay now that we know AMA's up and running, we're going to switch back. We're going to spin up the back end again. Python Devy. It already has the API endpoint for AMA local right there, which is perfect. Then up in the top right, I'm going to click the model drop down. We're going to scroll down. I'm going to select O Lama. So I had to refresh the page. I'm going to select a project new project Game of Life 2. Hit okay. Search engine Google. Select a model. Now we see MISTL is available under the O Lama tab. So let's select that. And I'm going to type create The Game of Life using Python with a UI. And then hit enter. And so this should be running completely locally now. Yep. And there it is. Awesome. So this is completely local. All right. Browsing the web. Let's see if this actually works though. All right. So I did actually get that Playwright issue again. And now that I'm thinking about it, I actually think it has to do with searching Google.

possibly, but it did find a Reddit link for T hinter. So it did do the Google search correctly. But when it got down down to actually writing it again with the browser, it failed. But local models now work perfectly. So you can run this completely locally. So again, it's a little bit buggy, but they're making great progress. Thank you so much to the Deva team and everybody who's contributing for building another awesome project. I can't wait to see where AI coding assistants go. The rate of maturing is just super impressive. If you liked this video, please consider giving a like and subscribe, and I'll see you in the next one.

Build ENTIRE Apps With A Single Prompt - FREE Open-Source Devika Tutorial

Artificial Intelligence (AI), Open Source, Generative Art, AI Art, Futurism, ChatGPT, Large Language Models (LLM), Machine Learning, Technology, Coding, Tutorials, AI News, and more

** Exclusive Pinecone Partner **

Matthew Berman

The Development and Progress of the DEA AI Coding Assistant

DEA is a fully-featured AI coding assistant with a rapidly evolving interface, offering support for various models and search engines.

The project has made significant progress in a short time, now allowing users to build applications quickly and easily, despite some remaining bugs.

The DEA AI coding assistant supports a wide range of features, including the ability to select different models and search engines, and it has the potential to simplify and accelerate the app development process.

Cloning, Setting Up Python Environment, and Installing Playwright for Deva

Clone the Deva repository from GitHub and create a new Python environment using conda.

Install the necessary requirements using pip and then proceed to install Playwright.

After that, set up the UI and run the development server.

Finally, spin up the backend by running a Python file, and wait for the Sentence Transformer model to load.

Setting Up Deva with Grok Mixol for Fast Powering

Deva is successfully up and running, using the Command and Local Host button.

The environment variable needs to be filled out in VS Code, including the Grok API key and the Google search engine ID.

The next step is to create a new search engine in the programmable search engine control panel.

Setting up Google Search Engine ID for DEA

Copy the search engine ID from the search engine and paste it into the VSS code for the Google search engine ID.

Then save the changes and restart the servers.

py" to quickly spin up the servers, then activate DEA again by navigating to the UI folder and running "Bun Run" to start the front end.

Encountered a Playwright Sync API error while trying to use Grok Mixol, leading to the decision to switch to ChatGPT, which resolved the issue and allowed the Game of Life project to work successfully.

The project quickly sets up the Game of Life and demonstrates its functionality.

The developer easily installs the requirements and runs the project, showcasing a simple and successful demo.

The process also involves setting up the project locally using Olama.

The user switched the back end to Python Devy and selected the O Lama model for a new project called Game of Life 2.

They experienced some issues with the Playwright and Google search, but overall, the local models are working well for them.