In my interview, OpenAI's CTO of Applications Vijaye Raji said more code gets thrown away now that coding is cheap. How to decide what your team keeps and scraps.
What OpenAI’s repository problem reveals about testing, review, and decisions—and how to identify the constraints in your own work.
TL;DR
- OpenAI's coding agents increased development work, straining existing infrastructure.
- Pando was built to address the code repository bottleneck caused by agent activity.
- The acceleration of coding led to delays in testing, review, and decision-making.
- FurtherAI implemented automated checks to handle repeated review comments.
- Vijaye Raji emphasized the importance of not getting attached to code and the value of throwaway prototypes.