404media.co web signal

Amazon Employees Mock AI Tool Kiro as 'Slop' in Internal Slack

TL;DR

  • Amazon staff nicknamed their AI-tool mockery 'Sloppenheimer' in an internal Slack channel called #actual-aws-memes, targeting Kiro and other AI products.
  • Amazon shut down its KiroRank leaderboard after employees gamed it by submitting pointless tasks, a practice they called 'tokenmaxxing.'
  • Amazon replaced KiroRank with 'normalized deployments,' measuring useful code output rather than raw token consumption.

Amazon engineers have been circulating memes about the company's AI coding tools in an internal Slack channel called #actual-aws-memes, according to 404 Media. Employees centered the mockery on Amazon's Kiro platform, dubbing their collection "Sloppenheimer," a reference to the 2023 film Oppenheimer, and using it to joke about AI-generated output they described as "slop." One meme showed Kiro's logo on a departing aircraft with the caption "Now I have everything I need," cut to stranded passengers and the line, "Narrator: He did not have everything he needed."

Workers said the campaign intensified around late 2024. As one anonymous employee explained: "Of course it doesn't help that leadership is definitely pushing AI so there's probably some element of backlash." That pressure took a concrete form in KiroRank, an internal dashboard tracking token consumption among developers using Kiro and Meshclaw. When the leaderboard was shut down, Amazon said it "achieved its goal of motivating staff." Workers alleged something different: that colleagues had been submitting pointless tasks to climb the rankings, a practice they called "tokenmaxxing." Senior VP Dave Treadwell reportedly told staff, "Please don't use AI just for the sake of using AI."

Amazon has since replaced KiroRank with "normalized deployments," measuring whether AI-generated code produces useful output rather than raw token consumption. The company defended Kiro as having "a fundamentally different approach to AI-assisted development that prioritizes production readiness and correctness," while noting the critical memes represented only a small sample of perspectives.

The honest caveat is that Slack meme channels are not a representative product review, and skeptical engineers in a single channel do not settle whether a tool is working more broadly. What the reporting does not fully answer is how "normalized deployments" will be calculated, or whether it will prove any harder to game than the metric it replaced. According to MLQ News, Meta runs a similar dashboard called "Claudeonomics" ranking workers by token consumption, suggesting Amazon is not alone in this problem and not the only company that may need to rethink how it measures AI adoption.

Shared on Bluesky by 6 AI experts (top 5 by trust)