AWS re:Invent 2025 - Crowning the Kiro code champions (DVT221)
Source: Dev.to
Overview
AWS re:Invent 2025 – Crowning the Kiro code champions (DVT221)
In this session Brian Beach celebrates the winners of Kiro’s Labyrinth coding challenge and introduces Kiro’s spec‑driven development approach. He explains how Kiro evolved from CodeWhisperer’s autocomplete to an agentic IDE, emphasizing spec‑driven development over “vibe coding.”
The challenge required participants to build a Python navigator that escapes text‑based mazes using depth‑first‑search algorithms. Kiro generated requirements, technical designs, and tasks before any code was written.
Winners
- 1️⃣ Paul – 1,314 turns (first place)
- 2️⃣ Robin – 1,450 turns (second place)
- 3️⃣ Thanh – 1,484 turns (third place)
Beach demonstrates how Kiro built the entire challenge application through its specification‑driven workflow.
This article is auto‑generated from the original presentation. Typos or minor inaccuracies may be present.
Introduction to Kiro and Spec‑Driven Development: From Autocomplete to Agentic AI
All right, hello. I’m Brian Beach, tech lead for the Developer Experience team at Amazon. I work on developer products such as CI/CD, CDK, CloudFormation, and now Kiro and other agent‑related services.
Session outline
- Quick introduction to Kiro and spec‑driven development (the focus of the coding challenge).
- Overview of the coding challenge itself, with a walkthrough example.
- Crown the top three winners and hand out prizes.
What is Kiro?
Kiro started as CodeWhisperer—an autocomplete tool that suggested the next line of code. Over time it evolved:
| Year | Milestone |
|---|---|
| 2023 | Autocomplete (CodeWhisperer) |
| 2024 | Chat‑based assistance, architecture discussions, and richer interactions |
| 2025 | Full‑blown agentic AI: can read documentation, query databases for schemas, run unit tests, and make informed coding decisions |
Kiro is positioned as the IDE for prototype‑to‑production, leveraging these agentic capabilities to enable a new development experience.
Spec‑Driven Development
The coding challenge highlighted spec‑driven development:
- Spec generation – Kiro creates a clear specification before any code is written.
- Design – Technical designs and task breakdowns are produced automatically.
- Implementation – Code is generated to satisfy the spec, with continuous verification.
This contrasts with “vibe coding,” where developers give a vague prompt and hope the AI produces useful code. Spec‑driven development reduces reliance on prompt engineering and improves predictability.
Video thumbnails (click to watch)
