As we mentioned, the Narrativa Generative AI platform can be broken down into four parts: Data, Recipes, Projects, and the CMS. We start with raw data and end up with automated text-based content and images.

First and foremost, we want to make it clear that the process of developing content is unidirectional. This means you should consider each of these four parts as steps in a process–starting with Data and finishing with the CMS. Of course, we can make go back and forth to make adjustments and modifications, but it's important to understand that a Project is not possible without Data. Conversely, Data that exist, but aren't attached to a Project won't allow you to finish your content.

This is the workflow that users typically follow on the platform:

• Data: We need to connect data to the platform, which must be in the form of a table. Each row in the table will generate a narrative specific to that row only. But don't worry about configuring that now; we'll handle this in a future module. It's important here to ensure that the data contains all the information we need.
  

• Recipes: As you make each AI recipe (organized as a new, additional column to the table), you'll create instructions that will be scaled across the entire dataset row by row. This allows you to originate the content that you'll integrate into your Project.



• Projects: Each previously configured row serves as an input in the content generation process. In this step, you will position blocks of text (conveniently called "Content Blocks") in the order you want them to be rendered in the final narrative. You'll also have the opportunity to pre-define several levels of variability, allowing the system to create unique and personalized narratives based on the provided data.



• CMS: Content blocks from the Project are compiled together to create a seamless composite narrative. The narratives (again, one for each row) are organized into a feed inside the Content Management System (CMS) to be reviewed and/or published. The feed can then be integrated your endpoint via an API URL.