A new player just entered the AI coding IDE market, and it’s Amazon. Kiro launched internationally on 7 May 2026 as a full replacement for Amazon Q Developer, which stopped accepting new signups on 15 May. If you’ve been tracking AI coding tools — Cursor, Claude Code, GitHub Copilot, Devin Desktop — Kiro is now a fifth name on that list.
What makes it different
Kiro’s central claim is spec-driven development. Where other AI coding tools work like this — describe what you want, get code, iterate — Kiro inserts a structured planning phase first. Before writing a line of code, Kiro produces three documents: requirements.md (using formal EARS notation), design.md, and tasks.md. The code is then generated to satisfy the spec, not the other way around.
This matters because the failure modes of pure prompt-driven AI coding are increasingly well-documented: intent drift (the agent misinterprets underspecified requirements), context decay (agents forget earlier decisions as the codebase grows), and unverifiable output (no objective way to check correctness). Spec-driven development is a direct response to all three.
Kiro also offers a standard “Vibe Mode” for quick tasks, and Agent Steering files — persistent instructions about coding conventions, naming standards, and architecture — that the agent carries across every session.
Where it stands
Kiro attracted 250,000 users in its first three months. At AWS Summit NYC on 17 June, Amazon launched Kiro Pro Max at $100/month and a native iOS app, suggesting serious investment in the platform. Under the hood, Kiro is powered by Claude via Amazon Bedrock.
What this signals
The AI coding tools market now has five serious platforms competing across spec-driven, agentic, and interactive workflows. For development teams, that’s optionality. For founders and product leads commissioning software, the more important signal is what’s converging: structured planning before code generation is becoming the professional standard, not an optional add-on.
Whether you’re evaluating a development partner or choosing tools for your in-house team, ask whether they’re working with spec-driven workflows. The gap between teams that are and aren’t is widening.
See how we apply AI-assisted development on client projects: AI-assisted development.