Amazon Opens Claude and Codex Access After Kiro Falls Short
Amazon lets employees use Anthropic’s Claude and OpenAI’s Codex, reversing a ban on new third‑party AI tools amid Kiro shortcomings.

Three images of Amazon prime boxes stacked with products from the boxes.
TL;DR
Amazon is granting staff access to Anthropic’s Claude and OpenAI’s Codex after acknowledging that its internal Kiro tool does not meet needs, despite earlier saying it would not support additional third‑party AI development tools.
Context
In November Amazon leaders issued an internal memo urging engineers to rely on its homegrown Kiro coding assistant over external options. The memo said the company would not plan to support new third‑party AI tools while continuing to back those already in use. Amazon has invested billions in rivals like Anthropic and OpenAI, yet its own AI coding tool has lagged behind, contributing to recent outages tied to AI‑generated code.
Key Facts
Amazon VP Jim Haughwout announced the expansion of agentic AI tools for employees to help them innovate for customers. An Amazon spokesperson noted that 83 % of engineers primarily use the in‑house Kiro coding tool. Both Claude and Codex will be made available through Amazon Bedrock, the company’s managed service for secure access to frontier AI models.
What It Means
The shift signals employee pressure overcoming a strict internal policy, highlighting gaps between Amazon’s AI investments and the usability of its proprietary tools. By offering Claude and Codex on Bedrock, Amazon aims to balance security concerns with developer demand while still promoting its own platform. The change may affect productivity metrics and influence future decisions about tool support.
Watch next: adoption rates of Claude and Codex among Amazon engineers and any impact on project timelines or internal tooling strategy.
Continue reading
More in this thread
Conversation
Reader notes
Loading comments...