Amazon Web Services just dropped three AI agents that could reshape how developers work. The standout is Kiro, an autonomous coding agent that learns your team's style and can work independently for days without human intervention. Announced at AWS re:Invent, these "frontier agents" handle coding, security reviews, and DevOps - representing Amazon's biggest bet yet on AI replacing routine development work.
Amazon Web Services just fired the latest shot in the AI coding wars. At AWS re:Invent Tuesday, the cloud giant unveiled three "frontier agents" that promise to handle everything from writing production code to running security reviews - with minimal human oversight.
The crown jewel is Kiro autonomous agent, which Amazon claims can work independently for days at a time. That's a bold promise in a market where most AI coding tools still require constant developer supervision. "You simply assign a complex task from the backlog and it independently figures out how to get that work done," AWS CEO Matt Garman told the re:Invent audience, according to TechCrunch coverage.
Kiro builds on Amazon's existing coding tool announced in July, but with a crucial upgrade - it maintains "persistent context across sessions." Translation: it won't forget what it's supposed to do halfway through a multi-day project. Garman described scenarios where Kiro could update critical code across 15 different corporate applications in a single assignment.
The agent learns by watching how development teams actually work. It scans existing code, absorbs coding standards, and builds what Amazon calls "spec-driven development" profiles. "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," Garman explained during his keynote.
But Amazon isn't stopping at coding. The AWS Security Agent works alongside Kiro to identify security vulnerabilities as code gets written, then suggests fixes automatically. The DevOps Agent completes the trilogy, handling performance testing and compatibility checks before code goes live. It's a comprehensive automation suite that could eliminate entire categories of routine development work.
The timing puts Amazon in direct competition with OpenAI, which announced last month that its GPT-5.1-Codex-Max model can run for up to 24 hours continuously. Both companies are racing to solve what developers call the "context window" problem - AI tools that lose track of complex, long-running tasks.
