How Ramp Uses AI in Software Engineering

It’s been several months since this blog was published: https://www.anthropic.com/customers/ramp

Given our partnership with Anthropic, and the fact that Claude Code is easily the best product out there for coding today, I get asked often on how to get started with AI in software engineering - beyond basic “code completion in IDE”. In response, I must have forwarded this blog several dozen times already.

And yet, it’s a relatively small number of organizations that have adopted platforms such as Claude Code for this level of automation. Easiest thing to do starting today - put these on a score card and start evaluating on a daily basis how you are ramping towards the speed and agility that Ramp has achieved (pun fully intended 😂)!

Here’s how Ramp engineers use Claude Code to move even faster:

1. Test automation: Ramp has developed extensions to Claude Code that streamline the development-test loop by connecting Claude Code to their testing frameworks through the flexible command-line interface. Since Claude Code can be invoked through the terminal and process standard input/output streams—similar to traditional developer tools—Ramp engineers can easily pipe inputs and outputs between it and their testing frameworks, creating custom abstractions that automate the repetitive cycle of writing code, creating tests, and fixing errors. Claude Code independently analyzes test failures, makes the necessary code adjustments, and reruns tests until everything passes—all through familiar interfaces that integrate smoothly with their existing development environment.

2. Enhanced documentation: Claude Code automatically generates comprehensive documentation across Ramp's codebase, improving the documentation process with consistent, contextual information useful for both current and future development.

3. Parallel development workflows: Developers run multiple Claude Code sessions simultaneously on the same codebase, each handling different tasks. This parallel processing eliminates wait time and significantly increases throughput.

4. AI-powered incident response: Ramp is building specialized tools for incident management by connecting Claude Code to their observability stack such as Datadog and Sentry. Claude Code integrates with Model Context Protocol (MCP) servers to autonomously aggregate logs, error reports, and system metrics when incidents occur, giving engineers consolidated, relevant information quickly. Early observations suggest this integration could reduce initial incident triage time by up to 80% because on-call engineers can quickly understand complex system issues without manually piecing together data from multiple sources.

5. Ticket-to-code automation: Engineers connect project management systems directly to Claude Code, which pulls ticket context, understands requirements, and implements solutions without manual information transfers.

Here’s to hoping we are not too far from when this will be table stakes and the team at Ramp will be publishing their next incredible blog on how they are moving even faster with AI!