On 11 November 2026, Amazon announced the AWS Emulator, a tool that lets developers run AWS services on their own machines. What makes it unusual is how it was built: according to Amazon, every line of code was written by thousands of autonomous AI agents, and no human wrote or reviewed any of it.

The emulator is open source on GitHub under the CC0 licence and is updated continuously.

How the agents worked

The work was split between three teams of agents:

• One team writes tests and checks them against the real AWS services.
• A second team runs the same tests on the emulator and records every failure.
• A third team fixes the failures, with each agent taking one issue.

The project started as an empty web server that accepted requests for 438 AWS API models, and it has since grown to tens of thousands of test cases. Before any change is merged, an automatic gate checks the build, coding standards, security, and any regression in how closely the emulator matches real AWS. Hundreds of agents work at the same time, and thousands of issues are resolved every day.

Why it matters

A local emulator lets developers test cloud applications without paying for real resources or waiting on the network. The bigger story is the method: the real services act as the reference, and tests decide what is correct, which is what allowed the agents to work without human review.

Source: About Amazon

By the TechMojaz team