How AI Agents Threw the Tech World Into Chaos

“Hi, my name is Peter, and I’m a Claudeholic.”

That confession came in August 2025, when Peter Steinberger stood before a meetup in London called Claude Code Anonymous. He and a roomful of fellow obsessives had gathered to connect with others swept up in the frenzy around Anthropic’s groundbreaking coding tool, Claude Code. “I dedicate pretty much all my waking time to this, yet it doesn’t feel enough,” Steinberger told the group, seated in a warm, brick-walled space.

Just months later, Anthropic unleashed a new version of Claude Code, and the ranks of Claudeholics swelled dramatically. Dubbed Opus 4.5, this iteration could tackle far more complex programming tasks, retain vastly more information in its memory, run for hours uninterrupted, and even orchestrate a team of AI subagents. Anthropic has a notoriously difficult take-home exam for prospective engineering hires. In a direct comparison between human candidates and its models, the company claimed Opus 4.5 “scored higher than any human candidate ever,” a result that “raises questions on how AI will change engineering as a profession.”

Throughout the holidays, countless coders retreated to basements and dens, manically experimenting with this new tool that let them build software as if they had unleashed a hundred clones of themselves or unlocked superhuman abilities. “It feels like becoming Spider-Man,” one enthusiast told me.

For Steinberger, now 39 and splitting time between homes in London and Vienna, even that wasn’t enough. In November 2025, he launched a project now called OpenClaw, a straightforward way to summon a personal AI agent that harnesses the power of Claude Code and similar tools. Grant it access to your data, your apps, and even your credit card, and it will comb through your cloud storage and venture onto the web to carry out your commands. It runs autonomously in the background, overcoming obstacles with the relentless persistence of the Terminator.

Steinberger’s creation took off by midwinter. One clear sign of its popularity: the number of “stars” a code repository earns on GitHub. In under two weeks, as users downloaded it and began feverishly building, the project racked up more than 100,000 stars. (By early May, it had reached 366,000 stars.)

With these two breakthroughs,the commercial product Claude Code and the open-source OpenClaw,the long-awaited era of AI agents has suddenly arrived. At least for those who are technically skilled enough, and perhaps bold enough, to dive headfirst into a messy, imperfect, and risky adventure. More than one Claudeholic has told me they feel they are living in the future. “AGI is here!” one fanatic declared, paraphrasing William Gibson’s famous line. “It’s just not evenly distributed.”

During the 1980s computer revolution, the general public regarded new machines with a mix of curiosity and anxiety while hackers joyfully built. A similar dynamic is unfolding today, possibly with even higher stakes. “It’s hard to explain how much of a sea change this is,” says Thomas Reardon, a former executive at Microsoft and Meta who now leads a startup in a different area of AI. “It’s the most underrated, massive release I’ve experienced in technology.”

Soon, we’ll all be experiencing it. On a recent podcast, Marc Andreessen,the co-inventor of the browser and self-proclaimed techno-optimist and MAGA supporter,made a proclamation that reflects Silicon Valley’s mindset: “It’s almost inevitable that this is the way people are going to use computers.” The unspoken corollary: It won’t be a choice.

Rewind to early 2024, when Boris Cherny was an Instagram tech lead, working remotely from a house he shared with his partner in rural Japan. “I would bike to the farmers market by the rice paddies,” Cherny, now 34, recalls. “Our hobby was making miso and pickles, and we would trade with our neighbors.” All of that changed when he began experimenting with the AI models emerging from his former hometown of San Francisco. (Cherny is originally from Ukraine; his grandfather programmed computers with punch cards.) The models jolted him from his peaceful life. Through friends, he connected with Anthropic, then moved back to the Bay Area to work there.