Welcome to building AI applications with haystack. But in partnership with Deep Set, the creator of haystack and taught by Tuana Çelik. This course will introduce a popular open source AI framework called haystack. Describe its components and syntax, and go on to show you how you could use it to build several AI applications. I often get questions from people wondering if they should use an AI framework, such as haystack, or just build an entire application from scratch. Building completely from scratch is a great learning experience, and also gives you full control over every step. By using a framework can help you get the job done faster, and also give you abstractions that make your code more maintainable and readable. Generative AI technology is evolving rapidly and it can be complex. Integrate APIs from different providers of LMS, better databases. There is to say, web search. And so on. And then there's also a lot of work to compose these. Too much analysis into custom workflow for your application. A framework like haystack can help manage this complexity, allow you to focus more on developing your application at a higher level of abstraction. For example, there are many vector database vendors, and using a common interface, your code can more easily adapt to changes in the underlying database technology without much refactoring. A few app frameworks can also provide common features out of the box, which can speed up your development process. For example. Hey, I suppose branching and looping. With branching, your pipeline can, for example, run a web search. If an initial retrieval step doesn't provide enough information to answer a user's query. Additionally, Insights Pipeline Visualization utility can help you to understand and optimize the process of building on workflows until later. There is chapter for this course is Tuana Çelik, who is Developer Relations lead for Haystack at Deep Set. Tuana has been helping many developers build the custom applications using haystack. Thanks, Andrew. That was a great summary of the reasons to use a framework. The points of haystack though, as a framework is not to provide everything that you may need, rather to provide a common interface and a simple abstraction with which you can extend the frameworks capabilities to your needs. For example, haystack is based on two main elements components and pipelines. It is built around the idea of having powerful components connected via flexible pipelines. We have many components like embeds and generators, but there are many cases in which haystack may not provide the components that you're after. For example, think of something that fetches data from a specific API. Adding this into haystack pipelines is a matter of creating your own custom component that can interact with this API, and haystack only expects you to wrap it up as a component. in the next hour of videos, you will learn about the unique building blocks that make up the haystack framework and then use them to build several exciting applications. You will learn about the core abstractions of haystack, including components, pipelines, and document stores, and see how these elements can be combined for various I use cases. You'll then move on to build and customize a simple right pipeline, and then learn how to tailor its behavior for your specific needs. Then you will learn how to create a custom component by building a. How can you summarize them? Additionally, you'll use conditional routing to create a branching pipeline with a fallback to web search when the context provided to the limb doesn't have the information needed to respond to the user's query. will also build a Self-reflecting agent using Haystacks Pipelines looping mechanism, enabling an agent to refine this response iteratively. Finally, you will create a chat agent that uses OpenAI's function calling capability, which allows you to provide hastag pipelines as tools to an enhancing that agent's capabilities. So that's a lot you learn in this course. Many people have worked to create this course from deep sets. I'd like to thank the entire haystack team, particularly Julian Risch Bilge Yücel and Madeesh Kannan. Also Esmaeil Gargari and Geoff Ladwig from Deeplearning.AI, have contributed to this course. I look forward to your building lots of exciting apps with haystack. Let's go on to the next video to dive in.