Ramp engineers now get substantive pull-request feedback in minutes instead of waiting hours, after adopting OpenAI’s Codex with GPT-5.5 across code review workflows.
Austin Ray, who leads AI Developer Experience, said the tool’s reasoning against Ramp’s codebase catches issues human reviewers and other AI reviewers miss, making it mandatory in many review flows.
Ray is also using Codex to build Ramp’s On-Call Assistant, an internal agentic tool meant to shoulder most of the burden during on-call rotations amid concurrency bugs, heavy incidents and complex business logic.
Ramp says the faster build cycle and higher confidence in shipped changes are reshaping engineering roles, with developers increasingly acting as orchestrators who direct, verify and iterate with AI tools.
How does Ramp prevent AI from introducing complex bugs that evade both AI and human reviewers?
Ramp accelerated coding with AI, but what is the new, unforeseen bottleneck in their development lifecycle?
With AI as the 'gold standard' reviewer, how does the crucial mentorship dynamic between engineers change?