Building Kodus and helping engineering teams ship with more quality and speed

substack

devto

Edvaldo Freitas  • Peerlist

AI changed how we write code. Now it’s time to change how we ship it.With copilots, writing became up to 10x faster — but shipping that code to production got 100x harder. Teams now face 8x more duplication, 60% of AI-generated code with quality issues, and even a 10% drop in overall system stability. That’s because today’s infrastructure for reviewing, testing, and shipping software was built in a pre-AI era, when speed and scale were completely different.Kodus exists to fix the path to production. We’re building the infrastructure layer for humans and AI agents to ship production-grade software.Our first step is Kody, an open-source code review agent that runs on GitHub, GitLab, Bitbucket, and Azure DevOps. Kody isn’t just another LLM wrapper: it understands repository context, learns from team-specific patterns, applies custom rules, and improves with every commit — powered by AST-level analysis and full repo history.The impact is clear:<ul><li>Senior devs save up to 3h/day previously lost to manual reviews.</li><li>First PR feedback drops from 12 hours to under 5 minutes.</li><li>Reviews become consistent, contextual, and adapted to your team.</li></ul>And this is just the beginning. Kodus is building agents across the entire lifecycle — review, testing, security, docs, and deploy — all embedded in your workflow. Our vision is simple: make shipping as easy as git push.