In August 2025, a meetup called “Claude Code Anonymous” was held in London, where organizer Peter Steinberger gathered a group of developers drawn to Anthropic’s Claude Code coding tool. The latest version, Opus 4.5, can handle more complex programming tasks, offers greater memory capacity, runs continuously for hours, and manages a team of AI sub-agents. Anthropic claims that in a notoriously difficult test designed to evaluate engineering job candidates, Opus 4.5 scored higher than all human candidates, sparking widespread discussion about how AI will transform engineering careers.
In November of the same year, Steinberger launched a tool called OpenClaw, which provides a simple way to create personal AI agents, leveraging advances in Claude Code or other coding tools. This breakthrough quickly triggered a chain reaction: Nvidia announced at its developer conference the adoption of a more secure version, NemoClaw; a wave of AI companies rushed to put agent capabilities into users’ hands. However, the autonomous multi-step execution of AI agents has completely upended traditional API billing models and forced companies to restructure their workforces at a fundamental level—independent programming subtasks are being compressed or eliminated, making skill transformation for developers not an option but an inevitability.
