Name: LLMStack
Uploaded: 2024-12-19T06:06:27.476Z
Duration: 2 min 33 s
Description: An open-source, no-code platform for building AI agents, workflows, and applications using your own data.

llmstack

An open-source, no-code platform for building AI agents, workflows, and applications using your own data.

LLMStack

Hi I'm a co-founder and CTO of Promly. In this video let me introduce you to LLM Stack, our open source no code platform for building LLM agents, workflows, and custom GPTs with your own data. 

For example, using LLM Stack you can build custom chat GYs like Chat BS that can answer questions from your own documents, EII agents that can browse the internet and generate research reports, and robotic process automation and many more. You can also call these apps or agents from other platforms like Flag, Discord, Toio, what etc. 

LLM Stack comes with everything one needs to build these AI applications. I'll show you how to install and configure LLM Stack, then quickly build an application and then give you a tour of the platform. 

LLM Stack can be installed using Python space, so before you install LLM Stack make sure you have Python.

3.10 Also make sure you have installed Docker if you're planning to use any of the advanced features like jobs or other um AI browser and things like that. So let me go ahead and install LM stack using this command. 

So I already have LM stack installed so it's pretty quick for me. So once you have install LM stag um you can run LM stag using the command LM stag. 

So if you're running this for the first time um it will ask you a few questions in order to create the initial configuration. Also download a Docker image that's needed for running the background jobs. 

So once the server is up and running it will take you to the login page. So you can log in into LM stack using U the credentials that you have provided during initial setup. 

So once you log in you'll be taken to the homepage where you can see all the apps app.

Templates and also the apps that you have created. Uh, on the sidebar you can navigate to various pages in LLM St. So let me, let us just go to settings and then I can show you the settings. 

So on the settings page you can configure keys for all the platforms that you're using. Here you can also like see the prompty token that you have here. So you will use this token or to use the API and trigger the apps that you have built on the platform or even manage the data programmatically in LLM stack. 

A procer is the basic building block. So all the apps that we talked about, they're all composed of various processors. Essentially, a processor chain will result in an app. 

So in playground you can find all the process that are available on the platform grouped by their providers. For example, here you can see the Open chat GPT.

Model is provided as a processor and here you can try out various processes and find a configuration that works for your application. Once you have the configuration that you are happy with, you can take that configuration and then bake that into your application. 

That you're building on the data page, you can import data into the platform. So once you have the data imported, you can use that data as a data source in your applications and then build various applications on top of that. 

So we now support uh importing data from various sources like URLs PDFs. We also support connecting to your own VV instance or even single store instance and then use those instances as a data source. 

And then so that you can connect to these things dynamically from your applications and then query them for data. And in the apps we have all.

These templates u in different categories and you can use these templates to quickly build applications. So templates will have the chain already built. So you only have to provide certain information that's needed for uh to change the behavior of the template and then create an instance for your own. 

Let me just add a data source and then we'll use the data source in our application. So let me create a data source called Lim stack uh from URL so let's use.

As the source and I can click this download uh button to list all the URLs from that page. And I'll only pick the URLs that I'm interested in for now. Okay, so let's I've selected a few um URLs and we can see uh we see that. As content is like ready now let's go to our app space and create an app. 

So for this Teo I'm going to use this website chatboard template which can answer users questions from the data source that that's provided. So I'll just click on the template right. Uh let's call it LM GPT. So this is essentially a custom GPT. 

Uh which can so this is the template page where I'm asked to fill these details. But then if you look in editor we can see the chain already built. It's already uh linked using this variable and then this is the output template for um this application. 

So let's go ahead and fill uh the.

Template or the data source whatever I want to use. Let's just use the newly created data source next. Okay, let's pick color and then let's save draft. So you can save an app as a draft while you're making the changes so that your publish tab doesn't get affected. 

So I can go to now preview and then say ask question like yeah we have our answer here. So I can look at the history and then see the question that I've just asked. Similarly, I can look at versions and then see the versions. 

So the app itself um can be configured using yaml which we'll talk in another video. But you can see the raw configuration that's actually behind the app. So now the app is working fine. Let me go ahead and save the app. 

So let's call it as first version. So once I've saved an app I can publish it. So let me just go ahead and publish the app.

I'll so we can select different visibility levels. I'll just mark it as public for now. Let me publish app and then we have a public URL that I can share there. So I can use this URL to share this app elsewhere. 

So now that we have published the app, we can embed this chatbot in various places. So I can use this code snippet and then embed this chatbot or this application in other web pages. Similarly, I can use the API to call this app. 

So this is the example code that I can just copy paste and we'll be able to call this app from other applications. So here is the Discord slack and too uh integration settings. So we can fill these things and create a Discord companion app. 

We'll be able to trigger this llm application from inside your Discord server. This is the overview of llm Stack.

Note, that there may be slight differences in text and video. Watch video for the original content.

LLMStack Video