Automating Code Review and Reducing Latency
Ramp’s AI Developer Experience (DevEx) team has integrated Codex with GPT-5.5 into their standard pull request (PR) workflows. By moving from manual reviews to AI-assisted feedback, the team has reduced the wait time for initial PR feedback from hours to minutes. The tool is noted for its ability to reason deeply against the codebase, catching complex issues that human reviewers often miss due to time constraints or oversight. The integration supports both CLI-based workflows for power users and a dedicated app interface that provides visual utilities, ensuring the tool fits into diverse developer preferences.
Building Agentic Internal Tools
Beyond code review, Ramp is leveraging Codex to build an "On-Call Assistant," an agentic tool designed to manage the high-complexity, high-context demands of on-call rotations. Managing incidents at Ramp involves navigating intricate business logic, concurrency challenges, and evolving incident details. Codex acts as a reasoning engine that helps engineers manage this cognitive load, allowing them to build and iterate on these internal tools faster and with higher confidence.
The Shift to AI Orchestration
Leadership at Ramp views the adoption of AI not as a replacement for human engineering, but as a fundamental shift in the developer role. Engineers are transitioning into "orchestrators" whose primary skill is directing AI tools, knowing when to trust model outputs, and when to intervene. To successfully implement these tools, the team emphasizes three leadership strategies:
- Hands-on Demonstration: Leaders should guide engineers through their first sessions to visualize the potential of the tool.
- Building Trust through Iteration: Treat AI integration as a learning process rather than an experimental toy to overcome initial skepticism.
- Direct Feedback Loops: Maintaining a direct line of communication with the vendor (OpenAI) is critical for resolving issues and ensuring the tool evolves to meet specific enterprise needs.