In August 2025, a meetup called "Claude Code Anonymous" took place in London. Organizer Peter Steinberger gathered a group of developers who had become deeply dependent on AI coding tools, sharing their "addictive" experiences with Anthropic's Claude Code Opus 4.5. This model could handle more complex programming tasks, boasted significantly expanded memory capacity, and could run continuously for hours while managing a team of multiple AI sub-agents. During internal testing, Anthropic compared it against human job candidates, claiming Opus 4.5 "scored higher than any human candidate"—a claim that ignited heated debates on how AI would reshape engineering careers.
In November of the same year, Steinberger released a tool called OpenClaw, allowing users to easily create personal AI agents that leverage Claude Code or other coding tools. Shortly after, NVIDIA announced during a major keynote that it would adopt a NemoClaw version based on OpenClaw, positioning it as a safer and error-reducing solution. This move was seen as a key milestone in bringing AI agent technology into the mainstream, prompting numerous AI companies to rush agent features to every user with a keyboard or smartphone. However, technological transitions often come with chaos, and the rapid deployment is surfacing issues around security, reliability, and ethics.
