Building an AI-native engineering org (twitter.com)

FurtherAI recently shared insights on transforming its engineering organization into an AI-native model, reporting a fivefold increase in code changes per engineer with a maintainable revert rate below half a percent. Key optimizations across the software shipping process have enabled this efficiency, including AI-written code and tests, automated reviews, and streamlined CI processes that run 20 times the tests in the same time frame. Notably, 60% of bug tickets are now filed automatically, and low-risk changes can merge without human intervention, allowing even non-engineering staff to contribute code effectively. As engineers become faster, they face new challenges related to decision-making and project planning. The focus is shifting towards empowering engineers to take ownership of their work from planning to execution, thus reducing bottlenecks and enhancing speed. This approach requires engineering managers to become more involved in the work and product-savvy, supporting engineers in making informed decisions without over-reliance on traditional roles. FurtherAI is now exploring how to maintain organizational agility at this heightened pace and is seeking to address the next set of challenges by hiring for key leadership positions.