Ramp Uses Codex to Cut Code Review Times
Ramp Codex code review workflows are now delivering pull request feedback in minutes rather than hours, according to a customer story OpenAI published on May 20, 2026. The fintech company is also using Codex with GPT-5.5 to develop an internal on-call assistant, marking a broader shift in how AI developer tools are being applied across engineering operations.
How Ramp Is Using Codex in Software Development
According to OpenAI's customer story, Ramp's AI Developer Experience team uses Codex with GPT-5.5 as part of its core software development process, with code review as the clearest example. The company says engineers now receive meaningful pull request feedback far faster than before, shrinking a process that previously took hours down to minutes.
Ramp says engineers rely on Codex because it reasons over the entire codebase rather than responding only to isolated prompts. That matters for review work, where understanding surrounding context is often as important as reading the changed lines themselves.
The tool also supports both CLI and app-based usage, which suggests it is embedded in regular engineering routines rather than deployed as a one-off experiment.
Why Faster Reviews and Agentic Tooling Matter
Faster feedback can accelerate more than a single pull request. When engineers receive review comments quickly, they can iterate sooner, catch issues earlier, and keep development moving without waiting on long review cycles. Ramp's experience adds to the growing case for GPT-5.5 code review tools built to reason across full repository context.
Ramp is also using Codex to help build an internal Ramp on-call assistant for engineers handling incident rotations. According to the announcement, that work is aimed at supporting engineers during on-call shifts, where systems, dependencies, and fast-moving incidents can be difficult to manage under pressure.
That positions this as more than a story about faster coding. It points to a wider pattern in how companies are deploying AI developer tools: not only to review and write software, but to build internal agentic systems that help teams operate production environments.
What This Signals for Engineering Teams
Austin Ray, who leads AI Developer Experience at Ramp, is described in OpenAI's customer story as seeing developers increasingly act as orchestrators of AI systems, directing tools like Codex across review, debugging, and operational tasks rather than writing every line manually.
[Analysis] Ramp's deployment offers a practical model for engineering teams evaluating Ramp Codex code review and agentic tooling. The path that emerges from this case: anchor AI tools in a high-frequency task like code review, build trust through consistent results, then expand into agentic use cases such as on-call support. Whether other organizations report similar gains in review quality and incident response will determine how quickly this pattern becomes standard practice.