Kiro - the AI IDE
Are you vibe coding anything? Well, if not then you should! the AI IDE’s are next level and with the recent new ones that caught my attention are really pretty solid!
First of all vibe coding is you guiding AI to build software and then you refine and debug it. The AI assisted coding is becoming better and better every single day with the models getting better. In simple words - Using English as your primary programming language is what defines vibe coding.
There are tools already available like Cursor, Windsurf and then CLI based too like claude code, gemini CLI, warp. All these tools are amazing and have improved developer productivity a lot. I recently came across Kiro - which is the latest AI IDE launched by AWS in direct competition with other tools. They are providing it for free in the preview beta stage and this is the first time I went ahead and added a feature to my existing application. Remember my Kubernetes course? NO?? Go check it out now, so in my Kubernetes course there were three games that you could have played, I installed Kiro and then asked if it could add another game. How it works?
It first generates a new spec with feature, requirements and acceptance criteria - exactly like you create a user story first and define all the steps.
This is just like a regular user story with requirements and acceptance criteria
Next is the design, once the spec is approved, it generates the design document. This doc defines the architecture, schemas, interfaces, integrations, data models error handling and gives a phase wise plan.
Last is the task list which is the actual implementation plan and that maps to the requirements too. After this it starts reading the code files and starts writing code.
Trust me, in one go, it was able to generate perfectly working new game for my application and did all the integrations in less than 5 minutes. This is so dope and the most I like is the user experience.
I am creating a video as well on this showcasing how cool it is, the prompts I used and its working along with MCP integration.
In short, Amazon's agentic AI IDE that's shaking up things in the AI IDE space. It's more than just an AI assistant; it's designed to be an intelligent partner that understands, acts, and reasons throughout your development workflow.
What makes Kiro stand out:
Spec-Driven Development: Kiro's core innovation is its "spec-driven development". Instead of jumping straight to code from a prompt, Kiro first translates your high-level ideas into clear requirements, a technical design, and a detailed task list. This means less "vibe coding" chaos and more structured, production-ready code. It even includes features like EARS (Easy Approach to Requirements Syntax) notation for acceptance criteria.
Agentic Autonomy: Kiro isn't just generating snippets; it's designed for autonomous, goal-driven actions. It can investigate your codebase, open relevant files, and modify them to fulfill your requests. This makes it feel like you're working with a junior developer who understands the goal and takes responsibility for solving the problem.
Agent Steering: A powerful feature is "Agent Steering". You can guide Kiro's behavior for your specific project by creating markdown files (product.md, tech.md, structure.md) that define your product's purpose, technology stack, and architectural conventions. This helps Kiro understand your project's unique context and institutional memory.
Agent Hooks: Kiro also offers "agent hooks" that automate development workflows. These are agentic actions triggered by file events (like creation, saving, or deletion) or manually. For example, you can set up hooks to automatically generate documentation, unit tests, or optimize code performance as you work.
MCP Integration: Kiro integrates with the Model Context Protocol (MCP) framework. This allows it to securely connect to external data sources and specialized tools, including AWS documentation servers, and potentially even your company's internal wikis or APIs, without exposing sensitive information to external AI services.
Control and Transparency: While Kiro can operate in "Autopilot Mode" for autonomous tasks, it also offers a "Supervised Mode" where it presents its plan and waits for your approval before making changes. You can review code diffs and execution history, maintaining full control.
VS Code Compatibility: Built on Code OSS, Kiro is compatible with your existing VS Code settings and Open VSX plugins, making the transition smooth.
Multimodal Input: Kiro can process diverse inputs, including files, codebases, images of UI designs or architecture diagrams, repository maps, Git diffs, terminal output, and even URLs.
Pricing: While Kiro is currently free during its preview period, Amazon has announced future pricing tiers:
Free: 50 agent interactions per month.
Pro: $19 per user per month for 1,000 interactions.
Pro+: $39 per user per month for 3,000 interactions. Additional interactions will cost $0.04 each. For Amazon Q Developer Pro users, Kiro will be free.
I feel that that Kiro can be slow, especially compared to Claude Code, and that the design/spec feature can over-engineer simple things. I think it would be good to have more model options beyond Sonnet, and better handling of CLI interactions on Windows (switching from PowerShell to Bash). However, the general sentiment is that Kiro is a solid first version and a significant step forward in AI-driven development.
If you're looking to bring more structure to your AI-assisted coding and bridge the gap from "vibe coding" to production-ready systems, Kiro is definitely worth exploring!
That was actually a complete blog in itself, I might put that up as a separate post :D
Coming to Kubesimplify - Thank you for the 20k views on the Kubernetes course, if you have not checked it out then you should!
FREE Kubernetes Certification voucher opportunity
I am excited to share that there is a opportunity to win a FREE certification voucher - you just need to fill the form and attend the livestream mentioned in the form.
Awesome Reads
Unlocking High-Performance AI/ML in Kubernetes with DRANet and RDMA - Google has announced the beta release of DraNet, a native Kubernetes integration that simplifies high-performance networking for AI/ML and HPC workloads by managing RDMA interfaces declaratively using the Dynamic Resource Allocation (DRA) API. DraNet reduces complexity, enhances performance, and improves security by enabling lightweight, standalone configuration of network interfaces directly within Kubernetes manifests eliminating the need for custom scripts or complex CNI chaining.
Grok 4, developed by xAI, is the most advanced AI model yet, featuring native tool use, real-time web and X search integration, and powerful reasoning via reinforcement learning scaled on a 200,000-GPU cluster. The Grok 4 Heavy variant sets new benchmarks across academic and competitive tasks, and introduces advanced voice and vision capabilities, while the Grok 4 API offers developers a 256K context window and enterprise-grade deployment options. I have not used it so cannot comment much but if you have used then do drop in comments.
Bare Metal Kubernetes with GPU: Challenges and Multi-Tenancy Solutions - In this post, I explain about the current AI infrastrcuture deployment and what are the challenges
Let's Encrypt rolls out free security certs for IP addresses - Let's Encrypt has started issuing free TLS/SSL certificates for IP addresses, a feature previously offered only by paid providers, enabling secure connections without requiring a domain name. While this caters to niche use cases—like hosting on static IPs or secure remote access—its six-day certificate lifespan and reliance on automation tools like Certbot reflect a push toward short-lived, more secure cert practices.
Ask ChatGPT
Under the hood: Amazon EKS ultra scale clusters - Amazon EKS now supports Kubernetes clusters with up to 100,000 nodes, enabling ultra-scale AI/ML workloads using up to 1.6 million AWS Trainium chips or 800,000 NVIDIA GPUs in a single cluster. Through innovations like offloading etcd consensus to a high-performance journal system, in-memory storage, API server optimizations, and enhanced node/network orchestration via Karpenter, AWS has drastically improved scalability, reliability, and performance for next-gen AI infrastructure.
Ask ChatGPT
Resources/Repositories
Learn from X
https://x.com/lexfridman/status/1944093274169323627
https://x.com/stevenbjohnson/status/1944803467085619631