Welcome to Build AI Apps with MCP Server, Working with Box Files, built in partnership with Box. In this course, you'll learn best practices for using MCP Server, illustrated by an example of processing unstructured PDF invoices to store extracted fields, such as the client name, invoice amount, and product names into a database. The invoices will be stored in a Box folder and you learn how to use the Box MCP Server to interact with files. I'm delighted that the instructor for this course is Ben Kus, who is CTO at Box. Thanks Andrew. I'm excited to work with you on this course. MCP or Model Context Protocol standardizes the way you provide context to an LLM-based application. Instead of writing custom integrations inside your AI applications to interact with external systems or data sources, you can connect your applications to MCP servers. and thereby automatically extend the set of tools available to your LLM. For example, Box's MCP Server provides a set of tools to interact with files and folders such as file search, text extraction, AI-based querying and data extraction for your data stored in Box. In this course, Ben will walk you through how to integrate the MCP server with any multi-agent AI application as well, in order to process your files. You use Box MCP as the example MCP server in the course, but the concepts are extendable to any other MCP server. You'll start by building a simple solution where you manually read each locally stored invoice and extract text from it. You'll then pass this text to your application's LLM to extract the client's name, total invoice amount, and product name from each invoice. Next, you'll streamline the process by using the Box MCP server. This server provides the tools you'll need to interact with and process invoices. As you add more features to your application, the logic may become more complex. So, the next step would be to transform your application into a multi-agent system that consists of multiple specialized agents, and where each agent can use the MCP server to search for or process files. For example, one agent can return the list of files inside a folder. And another agent can extract data from a given document. And maybe a third agent can generate a final report with the contents of the invoices. In this course, you will assume that these agents are running independently, maybe even developed by other teams. So you'll make them communicate with each other using the agent-to-agent or A2A protocol developed by Google. Many people have worked to create this course. I'd like to thank from Box, Scott Hurrey. And from DeepLearning.AI, Hawraa Salami and Esmaeil Gargari also contributed to this course. All right. Let's go to the next lesson. where you'll implement your AI application without the use of the Box MCP server, which requires that you write custom code for each file type.
Build AI Apps with MCP Server: Working with Box Files