This diagram illustrates a high-level overview of the generative AI workflow.
Using generative AI involves starting with a prompt. A prompt is a simple request that is sent to a language model to get a response. Prompt design is the process of writing a prompt to get the response you want from the model. There are prompt design principles and strategies that can help you guide the model to behave as you want. Prompt design may involve some trial and error.
Before being returned, the prompt and response go through a final layer of checks called safety filters. These filters check for toxicity in both the prompt and response, determining how much they belong to a safety category.
Prompts are sent to a model for response generation. Writer has a variety of generative AI foundation models that are accessible through an API or the Application layers
After the response is generated, Writer checks whether citations need to be included with the response. If a significant amount of the text in the response comes from a particular source, that source is added to the citation metadata in the response.
Once the prompt and response have been cleared by the safety filters, the response is sent back. Usually, the entire response is sent at once. However, it is possible to receive the response gradually by turning on the streaming option.
Updated 19 days ago