Code review powered by an LLM council (dromeas.ai)

A new AI-driven code review tool called Dromeas aims to revolutionize the way software development teams review changes. By leveraging a council of Large Language Models (LLMs), Dromeas provides multi-perspective analysis on pull requests (PRs), ensuring they are assessed for security, compliance, code quality, and functional intent. This approach addresses the growing challenges faced by developers, such as increased PR sizes and the complexity of AI-generated code, while fostering knowledge sharing and minimizing oversight. With key statistics showing a 98% increase in PR generation and a significant reduction in individual review time, Dromeas seeks to enhance efficiency without sacrificing thoroughness. The significance of Dromeas lies in its ability to streamline the review process at a time when developers are managing larger and more intricate PRs, often created with AI assistance. Dromeas flags discrepancies between PR descriptions and actual code changes to prevent surprises, while providing clear rationale for findings to promote learning within teams. Additionally, it offers customizable workflows that adjust review intensity based on the risk associated with each change. By addressing the visibility gaps between different teams—such as engineering, security, and compliance—Dromeas not only improves productivity but also ensures that critical aspects of code integrity are not overlooked in a rapidly evolving software landscape.