The rise of AI coding agents promised liberation from tedious coding, but the reality is a brittle software landscape. After a year, anecdotal evidence suggests 98% uptime is now the exception, not the norm, even for major services. User interfaces are plagued with bizarre bugs, and high-profile outages like the alleged AI-caused AWS incident reveal systemic fragility.

Companies boasting 100% AI-written code consistently ship low-quality software, undermining the promise of agentic development. Developers are finding themselves 'agentically coded into a corner' with no code review, delegated design, and a flood of unwanted features, leading to a loss of agency.