Giving agents a runtime environment to test their own code is revolutionizing software development, leading to unprecedented speed and innovation.

The New Stack is a publication covering trends and technologies in cloud-native development, DevOps, and software delivery. Developers can learn about containerization, Kubernetes, and cloud computing, as well as explore topics such as microservices architecture, serverless computing, and continuous integration/continuous delivery (CI/CD) pipelines.

The New Stack

Ramp's Inspect coding agent demonstrates that closed-loop AI agents with runtime environments are transforming software development. Unlike traditional open-loop agents that require human verification, Inspect autonomously generates, tests, and iterates on code until it works, accounting for 30% of Ramp's merged pull requests. The key innovation is giving agents access to sandbox environments where they can build, test, and verify their own code. This approach parallelizes solution discovery, reduces experimentation costs, and enables autonomous end-to-end testing. For complex cloud-native applications, connecting agents to shared baseline environments using Kubernetes infrastructure provides high-fidelity integration testing without replicating entire stacks.

Ramp’s Inspect shows closed-loop AI agents are software’s future