AWS Emulator and the engineer's job when agents write the code
Swami Sivasubramanian · 11 November 2026
Read the originalAWS built its new local emulator of more than 200 cloud services entirely with AI agents, and its engineers' work moved to the tests and merge checks that judge it.
Amazon Web Services (AWS), Amazon's cloud computing business, released AWS Emulator on 11 November, a free local copy of its services for testing software. In a guest post on the blog of Werner Vogels, Amazon's chief technology officer, Swami Sivasubramanian explains that AI agents wrote and tested all of its code. Engineers built the system that runs the agents and the checks that decide whether each change is accepted.
- Matt Goldberg, a Senior Principal Engineer on the AWS Developer Tools team, proposed that an AI agent could copy a service from its written description and the live service.
- The AI agents write end-to-end tests from AWS's public service descriptions and correct each until it passes against the real AWS cloud. They then build the emulator to pass them.
- Every change must pass a merge gate. Several security AI agents look for ways to attack it, a general code reviewer reads it, and fixed rules block changes to restricted code.
- Sivasubramanian once spent four nights supervising an AI agent on a rebuild before he saw it had no way to check its results. With a specification and a test environment, the rebuild took about two hours.
An engineering team that hands coding to AI agents can spend its own time on tests checked against a trusted reference and on the gate every change must pass. This is an inference, and the piece is the company's account of its own product launch.
Require every code change written by an AI agent to pass automated reviews, including security reviews, and fixed rules that block edits to restricted code before it can merge.
Have AI agents write end-to-end tests first and correct each one against the real system before they build the code that must pass it.
Derived by Working Surface from the article; more in the Playbook. Source line: Once it had a spec and a test environment, the rebuild took about two hours.
12 November 2026: when AI agents write the code, engineers' main work becomes writing and protecting the checks that decide whether that code is correct.