The future of automation at Atlassian: Generating Confluence automation rules with large language models

This post details the technical implementation of generating Confluence automation rules using LLMs, specifically OpenAI's models with function calling. It breaks down the process into three steps: Selection (trigger and non-trigger component selection), Generation (rule structure generation based on selected components and prompt), and Mapping (resolving references to internal representations). The post highlights the benefits of splitting the generation process for context limits and cost savings, and discusses strategies for handling user intent and error cases.