AWS Unveils Kiro: An AI Agent That Codes Autonomously for Days

Amazon Web Services (AWS) introduces three new AI agents, including Kiro, an autonomous coding agent that can work independently for days. These frontier agents, announced at AWS re:Invent, are designed to handle various tasks such as writing code, conducting security reviews, and automating DevOps processes.

New AI Agents from AWS

During his keynote at AWS re:Invent, CEO Matt Garman unveiled the three new AI agents. The most notable is Kiro, a software coding agent that can operate autonomously for extended periods. Kiro is built on AWS’s existing AI coding tool, also named Kiro, which was introduced in July. This new version of Kiro is designed to produce operational code by following a company’s software-coding specifications through a process called 'spec-driven development.'

“You simply assign a complex task from the backlog and it independently figures out how to get that work done,” Garman said. “It actually learns how you like to work, and it continues to deepen its understanding of your code and your products and the standards that your team follows over time.”

Key Features of Kiro

• Persistent Context: Kiro maintains context across sessions, ensuring it doesn’t forget what it was supposed to do.
• Autonomous Work: It can be assigned tasks and work independently for hours or days with minimal human intervention.
• Spec-Driven Development: Kiro learns and follows the coding standards and practices of the development team.

Other Frontier Agents

In addition to Kiro, AWS also introduced two other frontier agents: the AWS Security Agent and the DevOps Agent. The AWS Security Agent identifies and addresses security issues in code, while the DevOps Agent tests new code for performance and compatibility with other software, hardware, or cloud settings.

Industry Context

While Kiro and the other agents are not the first to claim long work windows, they represent a significant step forward in the automation of coding and DevOps tasks. For instance, OpenAI recently announced GPT-5.1-Codex-Max, which is also designed for long runs up to 24 hours. However, the effectiveness of these tools still hinges on their ability to maintain accuracy and avoid common issues like hallucinations, which often require developers to babysit the AI.

Implications for Developers

The introduction of Kiro and the other frontier agents has the potential to significantly reduce the workload for developers. By automating repetitive and time-consuming tasks, these agents can free up developers to focus on more complex and creative aspects of their work. However, the success of these tools will depend on their ability to consistently produce reliable and high-quality code.

References

• Amazon previews 3 AI agents, including 'Kiro' that can code on its own for days | TechCrunch