Ramp Boosts Dev Speed with Codex GPT-5.5

Ramp is accelerating its software development lifecycle by integrating OpenAI's Codex with GPT-5.5. This combination allows engineers to receive substantive code review feedback in minutes, a significant reduction from the hours previously required. This advancement streamlines the development process and enhances overall productivity.

The OpenAI News report highlights how Ramp's AI Developer Experience team uses Codex for critical code reviews. Austin Ray, leading AI DevEx at Ramp, describes the tool as the "industry gold standard," noting that engineers actively request its input on every pull request. This reliance underscores the tool's perceived value and its integration into mandatory code review workflows.

Codex with OpenAI GPT-5.5 distinguishes itself through its deep codebase reasoning, offering a thoroughness that often surpasses human reviewers' capacity. Ray emphasizes that the tool "meets engineers where they are," offering both CLI and app-based interfaces to suit different preferences. This adaptability ensures broad adoption and effective utilization across the engineering team.

Running Code Reviews That Teams Can Rely On

The AI Developer Experience team at Ramp is focused on boosting software development velocity and code quality through Codex. Ray states that Codex code review is an essential part of their workflow. Engineers who once faced multi-hour waits for initial feedback now get detailed comments from Codex within minutes.

Codex's ability to reason deeply against the codebase provides a level of detail that human reviewers, constrained by time, often cannot match. The user experience is tailored to engineers' needs, offering flexibility through CLI or a feature-rich application. Ray found the app particularly effective, describing it as a tool that "shepherds you toward higher productivity."

For complex coding challenges, Ray noted that Codex with OpenAI GPT-5.5 handles intricate issues with remarkable ease. This capability significantly reduces the mental overhead for engineers, allowing them to focus on innovation rather than getting bogged down in complex problem-solving.

Developing Internal Tools with Codex

Beyond code reviews, Ramp is employing Codex to build internal agentic tooling. A key project is the On-Call Assistant, designed to mitigate the burden on engineers during on-call rotations. Ray explains that on-call duties involve significant business logic, domain knowledge, and incident management, requiring engineers to maintain extensive context and navigate complex scenarios.

The development of the On-Call Assistant has been significantly accelerated by Codex's adept reasoning capabilities. Ray expresses increased confidence in the improvements shipped, noting that the tool handles Ramp's immense product surface area effortlessly.

Leadership Lessons

Ray advises leaders evaluating AI developer tools to focus on tangible impacts rather than mere demonstrations. He stresses the importance of hands-on experience and real-world results.

Leaders should encourage engineers to try Codex, guiding them through initial sessions to illustrate its potential. This approach helps build trust and encourages iteration, transforming experimental use into expert adoption.

A direct feedback loop with vendors like OpenAI is crucial for iterative improvement. Ramp's close collaboration with the Codex team has yielded significant progress.

"Codex is the real deal. Codex definitely helps us ship faster," Ray affirmed.

What’s Next

Codex is fundamentally changing how fast Ramp engineers can work, empowering them to pursue more ambitious projects. Ray envisions a future where engineers become orchestrators, directing AI tools like Codex rather than writing every line of code.

The core skill will shift to knowing how to leverage AI effectively, understanding when to trust these tools, and when to challenge them. Ramp's top engineers are reportedly adapting to this new paradigm the quickest.