Amazon instructs its AI coding assistant, Kiro, to be used in production, but about 1,500 employees want to use Claude Code
Amazon has announced that it will restrict internal use of Claude Code, an AI coding assistant developed by Anthropic, in favor of its in-house developed tool, Kiro. This decision has sparked criticism from engineers, who are concerned about slower workflows and reduced productivity, and it has been reported that internal conflicts have surfaced.
Amazon Engineers Revolt Over AI Tool Restrictions - TechRepublic
Amazon restricts use of Anthropic's Claude Code despite Microsoft urging tests - Storyboard18
https://www.storyboard18.com/digital/amazon-restricts-use-of-anthropics-claude-code-despite-microsoft-urging-tests-89486.htm
Amazon has invested a total of $8 billion in Anthropic, with its stake valued at more than $6 billion. Anthropic is also a major Amazon Web Services (AWS) customer, and has committed to using at least one million Trainium chips. The two companies have a close strategic partnership.
Amazon invests $4 billion into Claude developer Anthropic, co-developing AWS's proprietary AI chip - GIGAZINE
However, strict requirements are imposed on the use of Claude Code in internal production code development, and individual formal approval is required for use.
Kiro, the coding assistant Amazon recommends, is based on Anthropic's Claude model and integrates a suite of tools built by Amazon Web Services. According to a spokesperson, as of January 2026, approximately 70% of the company's software engineers had used Kiro at least once, demonstrating significant improvements in efficiency and speed of delivery. While Amazon supports existing tools, it has no plans to add support for other third-party tools at the time of writing.
This restriction has created an especially awkward situation for sales engineers who sell Claude Code to customers through Amazon's Bedrock platform. In internal forums, some employees have expressed concerns that it would be difficult to convince customers to use a product they haven't officially approved for production use, and that they need a free environment to support demos and builds. In internal discussions, approximately 1,500 employees have expressed support for officially adopting Claude Code internally.
Some engineers have pointed out that Kiro's performance is inferior to that of competing tools like Claude Code, hindering productivity, and some employees have warned that 'encouraging the use of our own tools through forced adoption rather than true value will actually hinder innovation and slow development.' There seems to be a shared sense of crisis among engineers that 'if our tools cannot keep up with our rivals, the original purpose of introducing AI to improve efficiency will not be achieved.'
Amazon has also been criticized for its lack of transparency in its internal decision-making process. Previous guidelines and internal messages stated that Claude Code had passed security and legal reviews, but employees have confirmed that this statement was later edited and removed. Many developers are growing increasingly distrustful of the current situation, as there is no sufficient explanation as to what data was used to reject Claude Code.
Related Posts:
in AI, Web Service, Posted by log1i_yk