New course: Functions, Tools and Agents with LangChain

Hi, I'm thrilled to be back with Harrison Chase, founder of LangChain and instructor on our new course Functions, Tools and Agents with LangChain. Hi Andrew, it's great to be back. In earlier courses we shared how to use LangChain, including how to use it to chat with your data. In the short time since we created those courses, there have been significant advancements in LLMs and the libraries the support they use as a developer tool. This course was created to update you on function calling, such as OpenAI's work on letting LLMs call other functions, which turns out to be very useful for handling structured data, something that LLMs has struggled with.

You've all seen the amazing ability of LLMs to work in natural language, but the rest of the computing world works with formatted data. Things like function calls or APIs want specific data in specific formats. With recent updates to training, LLMs can now understand and output data like JSON, and in this course you will get a chance to work with this directly. These updates make LLMs more predictable and reliable as well as being better at understanding when to use tools. This makes it more feasible to build agents that can reason about how to use tools to solve multi step problems. In this course we'll cover several things. First, we'll start by explaining some recent advancement in LLM APIs. Next, we'll go over a new syntax that we at LangChain have introduced called LangChain Expression Language, or lcel, which makes it much easier to compose and customize chains and agents. We will then take those two components and go over popular use cases like structured data extraction, function calling and building up to a conversational agent.

So that's a lot of very exciting topics. We hope you enjoy this course and learn how to take advantage of these new capabilities in your applications. We're excited to see what Yupo next.