Atlassian's Rovo Deep Research feature uses an enhanced Retrieval-Augmented Generation (RAG) architecture to generate comprehensive research reports from complex queries. The system breaks down tasks into manageable sub-tasks, performs parallel information retrieval, and creates structured reports with inline citations. It leverages multiple AI models including Llama 3.2, GPT-4o, and Claude 3.7 for different functions, and incorporates reasoning models for better planning and reflection. The feature integrates with Atlassian's Teamwork Graph to provide contextually relevant results while respecting user permissions.
Table of contents
How Deep Research improves knowledge search and report generationGenerating a contextual research planReflecting and refining resultsGenerating reports with RovoHow Deep Research leverages reasoning modelsHow Atlassian measures Deep Research qualityWhat’s next for Rovo’s Deep ResearchSort: