Ramp Engineers Accelerate Code Review with Codex

Ramp's AI DevEx team is using Codex and GPT-5.5 to accelerate code review while also building internal agentic tooling. Engineers now get PR feedback in minutes instead of hours.

Austin Ray says Codex code review catches issues that he, other engineers, and even other AI reviewers miss. Codex reasons deeply across the entire codebase to provide more thorough review, and it's used alongside both the CLI and the app to fit engineers' workflows.

Ramp is also building an On-Call Assistant with Codex. This tool helps reduce the burden of on-call operations by assisting with tasks like complex business logic, concurrency bugs, interactions between external and internal events, and long-running incident investigations.

Ray highlights three things for other leaders:

• Let engineers install Codex themselves and experience their first session
• Build early trust so it leads to repeated use
• Invest in a feedback loop with the Codex team

He believes that going forward, the engineer's role will shift from writing all the code themselves to being an orchestrator who knows when to direct AI tools and when to check them.