AIToday
Large Language ModelsAI Coding AssistantsZenn AI/MLPublished: Nov 14, 2026, 04:00 JST

Kiro and AI-DLC at Stage 4: agents may next find their own work

Send on LINE
Kiro and AI-DLC at Stage 4: agents may next find their own work

3 Key Points

  1. What happened

    The author traces Kiro from its July 2025 Agentic IDE launch through the December 2, 2025 Kiro autonomous agent preview and 2026 additions like Kiro Web, Automations, Memory and Model Fallback, and AI-DLC to its 2.11.3 stable release on November 13, 2026.

  2. Why it matters

    This progression shows agents moving from assisted coding to delegated work that persists across sessions, which raises the engineering burden of controlling, validating and governing them, in the author's view.

Not sure about something? Ask the AI

Questions and answers are published on this page.

Summaries like this, every morning.

Context & Analysis

Kiro's evolution is framed as a shift in the unit of work delegated to agents. Kiro launched on July 14, 2025 as an Agentic IDE, emphasizing structured development through Specs, Steering, Hooks and MCP rather than prompt-to-code generation. AI-DLC followed on July 31, 2025, addressing how the software development lifecycle itself should be redesigned for AI. At re:Invent 2025, the Kiro autonomous agent preview on December 2 and Kiro Powers the next day marked a move from working alongside an agent to delegating work to it.

Through 2026, Kiro Web, Automations, Cloud Sessions and Workflows expanded what could be delegated, while updates on October 12 and October 19 added Model Fallback, Session Reopen, Workflow Loop Recovery and Memory. AI-DLC added governance features and reached stable release 2.11.3 on November 13, 2026. The author's hands-on tests surfaced gaps: all validation tools passing did not guarantee a semantically correct change, and in CI/CD runs a one-line fix once became an 81-line change with a regression. A long-running AI-DLC workflow stalled repeatedly on a model traffic error, reported as GitHub issue #11521, which the author says shows the product's direction matched real problems but does not claim causation.

The author's central concern is that agent continuity is not the same as software lifecycle continuity, since state also lives in Git, GitHub, CI/CD, AWS and production. AWS DevOps Agent, generally available in March 2026 with a Custom SRE Agent added in June, hints at agents observing production and surfacing improvement candidates. The author speculates, clearly labeled as a guess, that agents may next discover the work to be done, shifting humans toward approval and policy roles.

FAQ
What is the next stage after Continuity & Governance?
The author suggests the next stage may be Proactive / Outcome-driven Engineering, where agents observe production signals and propose goals themselves, but stresses this is a hypothesis.
What is the difference between Kiro and AI-DLC?
Kiro is described as an execution environment that supports how agents develop, while AI-DLC is a lifecycle and workflow for how to delegate development to agents.
What evidence does the author cite from real usage?
In a seven-day Kiro Web Automations run on an AWS Serverless repository, detection was 18/18 against 18 ground truth items, and 14 of 16 auto-fix targets were handled without human correction.
Send on LINE

AI news that matters for your work, delivered every morning.

Pick your industry and the AI tools you use, and get news related to your work every day.

Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →

Ask AI

Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.

Questions and answers are published on this page.

Related Articles

Next articleAIエージェント can automate four HR and recruiting tasks, but people must decide