Overview and core features

0:00 But it can be difficult to learn. So that's why I'm making this video. I'm going to break it down for you as simple as possible. And by the end of this, you'll know how to build AI agents and you'll have a more complex one that you can adapt and change to your use case. Now, what we're going to do here is just quickly discuss the key differences between Langraph and something like Langchain. Then, we're going to hop onto the computer.

0:23 So, let's go ahead and do that. All right. So, let's scroll down here and just look at some stuff that they have on their website. So, Langraph really is meant for when you want to have a more complex application. It really is an orchestration framework. That means that it allows you to control the flow and kind of how your AI agent uses various state or passes through various stages. If you're not familiar, there are modules like Langchain.

0:45 Langchain is kind of a simpler, higher level framework that makes it easy to build simple AI agents, but when you have a lot of state or you have a lot of steps, it can be difficult to orchestrate that agent because it doesn't have those lower level features. So lang chain really just gives you much more control. It also is much better at persisting things like the context. So as you can see here, persisted context for long-term interactions.

1:09 So it knows kind of the information that you've given to it. And you can update and modify the state as you move through these various different steps. Now, as it says, lang graph uses a graph to represent data and to represent kind of the state of an agent. I'm going to show you that in just one second, but those are the main kind of key features. Now, of course, there are also advantages here when it comes to things like scalability and fault tolerance.

1:32 We're not really going to get into that in this video, but Langraph also has kind of this cloud platform that you can pay for that would allow you to deploy your Langraph applications. Again, we're not doing that in this video, but it is worth noting because that's kind of a huge selling point of using something like Langraph. Okay, so those are the main features. Now, what I want to do is I want to show you an example of a graph that you could generate with Langraph.

2:18 So have a look at these graphs here. On the right side, we have a very simple one. On the left side, we have one that's a little bit more complicated. Now, we're actually going to build both of these graphs. And these graphs will represent an agentic system or an AI agent. So, if you look at the right hand side here where we have start, chatbot, and then end. This is a super simple agent flow and something that you've probably built before without using something like Langraph.

2:18 Now, all this