Amazon Launches Kiro: An AI-Powered IDE to Streamline Vibe Coding

Amazon Kiro Is Here to Tame the Vibe Coding Chaos

Amazon Web Services (AWS) debuted a tool called Kiro today, which aims to formalize vibe coding, or a way to generate custom code through an AI chatbot. The vibe coding process typically entails talking to an AI until you are satisfied with its output. It's relatively unstructured, and not always effective. A recent study found it actually increased task completion time for experienced software engineers by 19%. Kiro offers a more professional approach that starts with project planning. Developers can enter specifications for each component of their project, and then vibe code to meet the requirements. "I'm sure you've been there: prompt, prompt, prompt, and you have a working application," says a blog post introducing Kiro, co-authored by Nikhil Swaminathan, AWS's head of agentic AI developer tools, and Deepak Singh, a VP at AWS. "It's fun and feels like magic. But getting it to production requires more. What assumptions did the model make when building it? You guided the agent throughout, but those decisions aren't documented. Requirements are fuzzy and you can't tell if the application meets them." Once you finalize the specifications, Kiro generates a technical design document based on your codebase. While you're completing the work, an AI agent works alongside you, acting like an "experienced developer" who catches your mistakes and completes routine tasks, Kiro says. It also performs quality checks once you've finished. The main AI model behind Kiro is Anthropic's code-focused Claude Sonnet 4, with Claude Sonnet 3.7 as a backup option, Amazon tells The New Stack. Support for more models is coming soon. Kiro is free for now while it's in preview. Eventually, the company will offer three pricing tiers: Free, Kiro Pro ($19-per-month), and Kiro Pro+ ($39-per-month). It will also be free for those with a Developer Pro account on Amazon Q, the company's AI coding assistant ($20-per-month). In a departure from other AWS launches, Kiro's website markets it as a standalone tool, with only a small AWS logo in the corner. Amazon has been working on it since at least May, when Business Insider first reported it. Amazon may still be deciding how it wants to sell the tool, whether it will be a part of AWS or exist independently. It will compete with other coding editors like Cursor and Windsurf. Google just hired the CEO of Windsurf for $2.4 billion after OpenAI's talks to acquire the company fell through, The New York Times reports. Microsoft GitHub's agent mode and Google Gemini Code Assist are also direct competitors from rival Big Tech firms, according to GeekWire.

AWS previews Kiro IDE for devs over vibe coding

Amazon Web Services has created what it's calling an "agentic IDE" that it claims avoids the pitfalls of vibe coding. With the advent of generative AI, developers have experimented with using LLMs to rapidly generate and debug code in a process that's come to be known as vibe coding. But the code is often of low quality and requires more time to debug and modify than it ultimately saves. A recent study found that in some cases developers believe these tools are saving them time when it's in fact precisely the opposite. AWS's tool is called "Kiro" and, as Deepak Singh, veep for developer agents and experiences, explained to The Register, it's built around a conversational interface that offers developers the chance to explain what they're trying to build. Kiro uses generative AI to produce its response, which initially takes the form of a spec, not actual code. Singh said the specs are "just markdown or text or pseudocode and are written like user stories." Nikhil Swaminathan, AWS senior manager for agentic AI developer tools, explained that each user story "includes EARS (Easy Approach to Requirements Syntax) notation acceptance criteria covering edge cases developers typically handle when building from basic user stories." Kiro's output also includes a list of services and actual code that's pushed to Git. Kiro can also handle multiple specs created by different teams that all work on different aspects of a project. Singh suggested that working with specs matters because when developers use coding assistants such as AWS's own Q, code quality is low and few bother to keep track of which prompts produce good results. AWS believes Kiro delivers code that's closer to production-ready, in an environment that's better suited to finishing a project and then maintaining it. To that latter end, Kiro offers event-driven automations called "hooks." Singh suggested Kiro users could create a hook that automatically reviews and optimizes code every time a developer adds code to a repository. AWS built Kiro on the open source Code OSS editor and can use plugins written for VS Code and Open VSX. Unusually for AWS, the product is a desktop client, but users can choose which cloud-hosted models it uses to generate specs and code. It's currently in preview, but AWS plans to charge $19.99 a month with a to-be-determined number of calls to LLMs. Singh said that despite having created Kiro and positioning it as an IDE for the agentic age, AWS still sees a role for coding assistants, and vibe coding as a way for developers to tinker and experiment. ®

Amazon jumps into AI vibe coding with preview of Kiro

Amazon CEO Andy Jassy attends the Allen & Company Sun Valley Conference on July 9, 2025, in Sun Valley, Idaho. Amazon's cloud unit said Monday that it has released a preview of Kiro, a program that developers can use to write code with help from artificial intelligence. In a post on X, Amazon CEO Andy Jassy said Kiro "has a chance to transform how developers build software." The introduction comes days after Google said it's hiring staffers of AI coding software startup Windsurf as part of a $2.4 billion technology licensing deal. Google said it plans to make its Gemini AI models more useful to software developers. Amazon and Google are jumping deeper into so-called vibe coding, the process of directing computers to create software with minimal human direction. Microsoft has also bolstered its Visual Studio Code editor with an agent mode for automated software development. Windsurf competes with Cursor, whose parent company Anysphere was reportedly in talks to raise money earlier this year at a $10 billion valuation. OpenAI looked at acquiring Windsurf and Cursor. Amazon Web Services, the leading provider of cloud infrastructure, said on the frequently asked questions page of the Kiro website that vibe coding in its current form can be overly complex. "When implementing a task with vibe coding, it's difficult to keep track of all the decisions that were made along the way, and document them for your team," the site says. "By using specs, Kiro works alongside you to define requirements, system design, and tasks to be implemented before writing any code." Kiro can make diagrams to show how data will flow through a proposed application, and create task lists, so that people can see what's missing, Nikhil Swaminathan, product lead at AWS, and Deepak Singh, the group's vice president of developer experience and agents, wrote in a blog post. Kiro currently can only chat with people in English. Support for additional languages will come later. The program draws on AI models from Amazon-backed Anthropic, but alternatives will follow, AWS said. Free and premium tiers of Kiro will be available after the preview ends. Content from paying users won't be used to train models, and free users can opt out, AWS said.

Amazon targets vibe-coding chaos with new 'Kiro' AI software development tool

A new AI coding tool from Amazon uses agents to automatically create and update project plans and technical blueprints, aiming to solve an increasingly common business headache: undocumented AI-written software that becomes difficult or impossible to maintain. The new tool, called "Kiro," (pronounced keer-oh) is an AI-driven integrated development environment, or IDE. It launched in preview Monday morning. The Amazon team behind the project says it's aiming to bridge the gap between rapid AI-generated software prototypes and production-ready systems that require formal specs, comprehensive testing, and ongoing documentation. The idea is to go from "vibe coding to viable code," as the Kiro website puts it. The move puts Amazon in direct competition with existing tools like Microsoft GitHub's agent mode and Google's Gemini Code Assist, as tech giants race to introduce AI assistants capable of handling complex software projects with minimal human oversight. Kiro emerged from a small team within Amazon Web Services, but in a departure from typical launches, it is hosted on its own domain, and Amazon's name appears nowhere in the announcement. Only an AWS logo and links in the website footer signal the connection between Kiro and the tech giant. It's part of Amazon's push beyond basic coding assistance into autonomous AI software development. While the Amazon Q Developer tool focuses on code completion and chat-based assistance, Kiro deploys AI agents that can act autonomously to help complete and document projects. The project was introduced in a blog post by Nikhil Swaminathan, Kiro's product lead, and Deepak Singh, Amazon's vice president of developer experience and agents. The vision, they write, is "to solve the fundamental challenges that make building software products so difficult -- from ensuring design alignment across teams and resolving conflicting requirements, to eliminating tech debt, bringing rigor to code reviews, and preserving institutional knowledge when senior engineers leave." As described in the post, Kiro works by breaking down developer prompts into structured components -- requirements, design documents, and task lists -- to guide implementation and testing. It also tracks changes and updates those materials as the code evolves, aiming to reduce inconsistencies between what's planned and what gets built. It uses automated checks that run when developers save or modify files, handling routine updates like refreshing documentation or scanning for potential problems. Kiro is free during its preview period. Ultimately, the company plans three pricing tiers: a free version with 50 agent interactions per month; a Pro tier at $19 per user per month with 1,000 interactions; and a Pro+ tier at $39 per user per month with 3,000 interactions. Business Insider first reported on Kiro in May based on a leaked document.

AWS Enters the Vibe Coding Market with Kiro AI | AIM

Kiro is positioned as an agentic IDE, an environment where AI agents collaborate throughout the software development lifecycle, not just at the prototyping stage. Most AI coding tools assist developers in getting started but struggle with the final output, as they often encounter documentation delays, missing tests, and broken integrations. AWS's new IDE, Kiro, launched on July 14 in free preview, aims to address these issues with a spec-driven workflow that supports developers from prompt to production. "It's fun and feels like magic to prompt your way to a working application," said Nikhil Swaminathan, product lead for agentic developer experience at AWS. "But getting it to production requires more." Kiro is positioned as an agentic IDE, an environment where AI agents collaborate throughout the software development lifecycle, not just at the prototyping stage. In an exclusive interview with AIM, Srini Iragavarapu, director of generative AI applications and developer experiences at AWS, said Kiro was built by a lean internal team that had been observing how AI-assisted coding evolved over time. The IDE supports Claude Sonnet 4 and 3.7 and includes agentic workflows, multimodal capabilities, and spec automation. Iragavarapu further added that one of the features the team had planned ahead of Kiro's launch was a notification system, something that would alert the user once an agent completed a task in the IDE, allowing them to focus on other work in the meantime. "That kind of feature would typically take us a few weeks to build across Windows, Mac, and Linux," Iragavarapu said. "But the team used Kiro itself to build it, within days." At the heart of Kiro is spec-driven development, where the IDE interprets the developer's intent and builds structured specs before generating code. Explaining how Kiro handles development tasks, Iragavarapu said the IDE first tries to understand the context of what the developer is trying to do. Instead of jumping straight into generating code, it starts by talking to the developer. "It will provide specifications of sorts, like a step-by-step task list," he said. Developers can then review and adjust these tasks before execution. Kiro then sequences the tasks based on dependencies and adds metadata such as unit tests, accessibility requirements, and loading states. Progress can be tracked through manual triggers, inline diffs, and agent execution logs. Developers can trigger tasks manually and monitor their progress, with inline code diffs and execution history available for review. "Specs are artifacts that prove useful anytime you need to think through a feature in-depth, refactor work, or understand system behaviour," said Swaminathan. These specs evolve with the codebase and stay in sync throughout development. Kiro also analyses the existing codebase to generate technical designs, including TypeScript interfaces, database schemas, API endpoints, and data flow diagrams, reducing ambiguity between design and implementation. Another unique feature in Kiro is agent hooks, which allow users to automate development workflows by monitoring source code for specific triggers, Iragavarapu explained. "It is always listening to your code changes, and when your code changes...it will automatically update your documentation." He explained that in a typical setup, developers often make frequent code changes, but the documentation doesn't get updated in real time. This leads to mismatches that need to be manually fixed later. Additionally, developers can define agent steering documents to set architectural or coding guidelines. For instance, Iragavarapu shared how he instructed Kiro to 'use React,' 'write TypeScript,' and follow a specific directory structure. The AI follows these rules throughout the development process. The tool is currently available in a free preview. Developers can sign in with GitHub, Google, or existing AWS IAM Identity Centre credentials. Pricing tiers will be announced when the product reaches general availability. Kiro currently supports Claude Sonnet 4 and 3.7, with options for developers to switch between the two. "More models are coming soon," confirmed Iragavarapu, who also noted the IDE's multimodal capabilities. Developers can upload hand-drawn architecture diagrams and have them converted into AWS CDK code. Kiro complements Amazon Q Developer. "If you already have a Q Developer subscription, you can also use Kiro," said Iragavarapu. Moreover, Kiro supports default MCP (Managed Context Provider) servers but also lets developers plug in custom ones, including internal or private sources. This means users can bring additional context into the IDE, whether from public or private MCP servers. Built on Code OSS, Kiro remains compatible with existing VS Code settings and Open VSX plugins. Inside Amazon, employees already use internal MCP servers to enhance development workflows."If you have your own favourites or your own private MCP servers, you can use them," Iragavarapu said. AWS is stepping into a crowded, highly competitive space. Google recently hired the founders of Windsurf in a $2.4 billion licensing deal. Tools like Cursor, GitHub Copilot, Replit, and Lovable have already built strong developer followings. When asked how they plan to challenge these, Iragavarapu said the goal is not to replace but to offer an alternative option. "We are providing options to developers. The way Kiro differentiates itself is through spec-driven development, advanced agent hooks, and agent steering, all from ideation to deployment." He clarified that while Copilot might assist in code generation, Kiro focuses on delivering production-ready features by automating and enforcing the full development lifecycle. AWS is treating the preview phase as a learning opportunity. "We'll share examples of what Kiro has built internally soon," Iragavarapu said, adding that the goal is to iterate quickly and evolve Kiro into a production-grade, AI-first development environment.

AWS launches Kiro: a 'spec coding' developer environment integrated with AI agents - SiliconANGLE

AWS launches Kiro: a 'spec coding' developer environment integrated with AI agents Amazon Web Services Inc. today launched a preview of a new development environment named Kiro, integrated with artificial intelligence agents for software engineers, which the company says will help them turn ideas into production-ready code. Now in preview, Kiro helps provide speed and resilience to has become known as "vibe coding," a new way to use development tools to tell an AI assistant what the developer wants built using conversational English and then working with it like a pair programmer or sitting back and letting it do most of the work. Amazon's newest tool is an integrated development environment, or an IDE, which is a software development interface where software engineers spend most of their time building, coding, testing and compiling software. Traditionally, the experience of vibe coding might start from a blank template or an existing app where a coder prompts the AI to write something. Then they prompt it again to either write more or fix what it wrote. This chain of prompts eventually leads to a final product. Amazon said that Kiro will change this with integrated AI agents that will build in "specs" and use "hooks" that will understand the width and breadth of taking a prototype to production. As a result, Amazon calls what Kiro's new capability "spec coding." For example, imagine a developer has an e-commerce app. For specifications, Kiro will take a single prompt, such as "Add a review system for products," and generate a requirements document and user stories for viewing, creating, filtering and rating product reviews. Each story will include acceptance criteria and edge cases. This can be turned into Kiro tasks and sub-tasks that the agents can then send to coding agents. Each task includes details such as requirements, implementation, accessibility and testing needs. This allows developers to follow along and check the work in steps to avoid any missing pieces. "Kiro's specs stay synced with your evolving codebase. Developers can author code to update specs or update specs to refresh tasks," AWS Product Lead Nikhil Swaminathan and Vice President of DevEx and Agents Deepak Singh wrote in a blog post. The important thing about this approach is that the code and the agent's process are completely documented top-to-bottom. Nothing is left out and the developer has a bird's eye view of how the app or function will be built and is able to guide it from the requirements view before anything happens. Amazon said this eliminates the costly back-and-forth usually associated with vibe coding. Once the developer hits go, they can watch the new code being built with an execution status indicator. Hooks work slightly differently from the specs; they're more like an experienced developer paying attention to changes in code. A hook triggers when a developer saves or creates a file, prompting an AI agent to review it and take necessary actions. For example, a developer saves a React component, and a hook has an AI agent update a related test file. When an application programming interface endpoint is changed, a hook updates the related README documentation. Upon preparing to deploy the software, hooks can run the software through a vulnerability scanner to make sure no passwords, API secrets or other credentials have been accidentally leaked. Hooks work to enforce best practices and other rules for the entire team, making sure that developers are keeping up with quality. They can go down checklists, update documentation and provide security validation as files are saved and updated behind the scenes. Amazon said Kiro goes beyond specs and hooks for developers and can be expanded with the Model Context Protocol, an open-source method for connecting AI agents to external tools. This provides developers with an extensive library of open-source AI tools they can connect to increase Kiro's coding potential. Of course, for developers who want to go line-by-line, Kiro also includes an agentic chat for coding tasks within a file. "The way humans and machines coordinate to build software is still messy and fragmented, but we're working to change that," the Kiro team wrote. "Specs is a major step in that direction."