# AWS Launches Emulator to Enable AI Agents for Distributed Systems New local emulator allows AI coding agents to test AWS services offline reliably. - **Source:** [Hacker News](https://news.ycombinator.com) - **Published:** 2026-11-11 - **Topic:** [technology](https://brifing.si/c/technology) - **Original article:** - **On Brifing:** https://brifing.si/s/1d01a7d1-91ad-419e-b48d-432f377159a6 ## Evolution of Distributed Testing Amazon CTO Werner Vogels reflects on the history of distributed systems, from the early Dynamo database to the Alf locking service. Two decades ago, engineers relied on JVM-based simulations and formal reasoning like TLA+ to ensure correctness. These manual methods were essential for managing complex Paxos-based replication across data centers. ## The Challenge of Scale Building reliable distributed systems requires handling network partitions, server failures, and massive concurrent client requests. Historically, testing these systems was expensive and difficult to reproduce on demand. While chaos testing and large-scale system tests were effective, they could not simulate every specific event ordering needed to catch subtle concurrency bugs. ## Introducing AWS Emulator AWS has released AWS Emulator, a tool providing a local, offline copy of AWS services for AI coding agents. It utilizes over 400 public API models in Smithy format to replicate service behavior. This allows agents to verify their code against a local environment before deploying to the actual cloud infrastructure. ## Agent-Driven Development The emulator was built and tested entirely by AI agents. Starting with a prompt for an SQS emulator, the system evolved to cover 200 services. Agents work in a continuous loop, writing code and running fidelity tests against real AWS to ensure their local emulator matches the production service's actual behavior. ## The Role of Fidelity Tests Fidelity tests serve as the primary guardrail for AI agents, defining expected outcomes based on public API models. If a test fails against real AWS, the agent identifies a gap in the emulator. This iterative process ensures that the local environment remains a high-fidelity representation of the complex cloud services. ## Redefining Engineering Tasks As agents take over coding tasks, the engineer's role shifts toward designing robust guardrails. This includes building specifications, defining tests, and creating merge gates. By providing agents with the right environment to check their own work, engineers can accelerate development cycles significantly, as seen in a recent replication engine rebuild. ## Security and Merge Gates Every change made by an agent must pass through a strict merge gate before entering the codebase. This automated process includes security agents that perform adversarial reviews to identify potential vulnerabilities. These deterministic checks ensure that the code produced by agents meets the high reliability standards required for AWS infrastructure. ## Open Source Commitment The AWS Emulator code is released under a CC0 public domain dedication, making it accessible to the broader developer community. The project's full commit history and issue tracking are available on GitHub. This transparency allows users to see exactly how agents identified and resolved discrepancies between the emulator and real AWS. ## Future of AI Engineering AWS expects agents to handle the majority of software construction, with engineers focusing on high-level architecture and verification. The development of AWS Emulator serves as a practical blueprint for this new paradigm. Future updates will continue to refine these guardrails, enhancing how teams leverage AI to build reliable, scalable software. ## Sources covering this story - https://www.allthingsdistributed.com/2026/11/aws-emulator-and-the-engineers-job-when-agents-write-the-code.html — --- _Summary written by Brifing from the original reporting. Copyright in the original article belongs to its publisher; read the full piece at the link above._