In this lesson, you will apply and see the power of the search and discovery approaches when using the AI agent to complete a couple of different tasks on a large production code base. Let's dive in. Great. In this demo, we're going to go and get another language in another repository just to show the breadth of what Cascade is capable of doing, specifically on large repositories, using some of the search and discovery tools that we just talked about. So this is a repository with a lot of Java code, but 100,000 lines or so has many files, it uses a lot of different frameworks and packages, things like Groovy and Spring Boot. And let's say I am a new developer that is onboarded onto this code base. This would normally take a while, but with some of those powerful search and discovery tools that we talked about, Cascade is able to help speed us up. So I'm going to go to Cascade, and I'm going to ask it the question: Explain what the code in my tag the entire source directory, some 300 odd files does. Let's see what Cascade does. So Cascade will use a number of different search and discovery tools that we discussed. First, it will use the LS tool to actually figure out what are all the top-level directories. And over here it did a riptide call. So as you can see, over almost 3000 different code snippets were searched upon and assessed for how relevant it is to the problem of understanding this code base. Then it went particularly to the test directory to understand a little bit there another riptide call. Again, you saw here multi-step retrieval. There was multiple retrieval calls before we even tried to give an answer. And here you go. It explained me a pretty concise manner. All of the notable pieces that I need to know about this repository. It even provides me some nice links into the code base for me to look there. Again, I can always follow up with questions. So let's say I'm going through here and I say, yeah, the book management is interesting. I want to learn more, explain me more how the book management happens. This is a conversation where the previous part of the trajectory and the previous part of the conversation is also incorporated, but as you can see, we will do another riptide call that is more looking at the relevance for this particular new query. So you can see we're doing multiple of these search and discovery tool calls. Before coming up with an answer to the question. And again, comprehensive explanation with tons of citations. Back into the code base that I can click on and read up and understand myself. So that's the power of these powerful search and discovery tools on real production code bases. But I'm going to show, yet another application of search and discovery in different kind of tool. Let's say I actually want to perform a migration on this. Let's do it on Spring Boot. I would ask Cascade what version a Spring Boot is being used, and what is the most recent version? The first half of this question is going to require understanding the code base, and the second half of the question will likely have to go out to the web. So it's going to do analysis. Figured out what the current Spring Boot version being used is. And now you can see that web search is happening. We're going out to the web cascade. It does a search for latest Spring Boot version. It then takes the results, creates an outline that I can actually both go and view the page. Or I can do in my editor exactly what the text is that it's analyzing, and it's giving me the information that wouldn't be present in the pre-trained model. And it's even telling me, hey, there's a significant version gap for which require multiple version changes, and it's going to start guiding me towards a migration process. I'm not going to go through the full migration, but this highlights in this demo, a number of different search and discovery tools. We saw the LS tool being used, Riptide being used, web search being used each time in a multi-step approach. And this is what allows agentic systems to address some of these more complex, long-running tasks. Hopefully, this is exciting to understand exactly how these powerful search and discovery tools can be used on a production code base. But let's switch gears and start building a new app to explore a lot more that Cascade has to offer.