The rush is on. Enterprises are increasingly turning to AI coding agents, and the tech giants are battling to be the platform of choice. In this crowded space, Amazon Web Services (AWS) is doubling down on a different approach with its Kiro coding agent.
Launched in public preview in July, Kiro is now generally available, boasting new features aimed at ensuring code behaves as expected. These include property-based testing and a command-line interface (CLI) for custom agent tailoring. The idea? To provide structure to the often chaotic world of AI-assisted code generation.
Deepak Singh, AWS vice president for developer agents and experiences, puts it this way: Kiro “keeps the fun” of coding while bringing much-needed order. It’s about enabling developers to collaborate with agents in a structured manner, converting ideas into enduring, robust code. Spec-driven development, as AWS calls it.
One of Kiro’s key differentiators is its focus on behavioral adherence. The challenge with AI-generated code, as many enterprises have discovered, lies in verifying its accuracy and ensuring it aligns with its intended purpose. Traditional testing can fall short, limited by human biases and missed edge cases. AI can even “game” the system, modifying tests instead of fixing code.
Property-based testing, a new feature in Kiro, tackles this head-on. It leverages specifications to identify desired code properties, automatically generating hundreds of test scenarios. This helps verify that the code behaves as intended, catching edge cases and ensuring implementation matches intent.
Imagine building a car sales app. A specification might state: “For any user and any car listing, WHEN the user adds the car to favorites, THE System SHALL display that car in their favorites list.” Property-based testing then automatically tests this with a myriad of combinations – User A adding Car #1, User B adding Car #500, users with special characters in usernames, cars with various statuses, and so on.
Kiro also introduces checkpointing, allowing developers to revert to previous changes if things go awry. The Kiro CLI, another new feature, brings the coding agent directly into the developer’s command-line interface, reducing context switching. This leverages functionalities from the Q Developer CLI, launched in October 2024.
The competition is fierce. From OpenAI’s GPT-Codex to Google’s Gemini CLI and Anthropic’s Claude Code, developers are demanding seamless access to coding agents within their existing workflows. Some platforms even allow users to choose which model to use for coding. Kiro, however, routes to the best model for the task, including AWS models, moving beyond reliance on a single LLM.
Singh notes that enterprises are recognizing the benefits of AI-powered coding, with brands like Monday.com already seeing significant gains. “We saw that the mental model changes for developers, but it’s not just about becoming more efficient; it’s also how they organize around the way they work now,” he said.
AWS is offering startups one year of free credits to Kiro Pro+ and expanded access to Teams in most countries. In a sea of coding agents, AWS is betting that structure and fidelity will win out.
About the Author
