Three scenes with Jean Phil.

If agents take on more implementation, the gain should be more than throughput. Here are the choices that could return judgment, learning, and time.

AI software factories now span human-reviewed agents and no-review designs. The hard question is who defines success, authority, and risk.

Software keeps getting ordered to act like a factory. Sixty years of hard lessons explain the values our practice should protect, whatever comes next.

Neuromorphic computing is leaving the lab. Chips that compute the way biology does are heading into consumer products, robotics, and smart homes, and they may be the only sustainable way to prevent AI infrastructure from devouring the electrical grid.

Ural survived eighty years of Soviet collapse, freezing mud, and bankruptcy, only to be strangled by Vladimir Putin’s imperial war. Its Chinese replacement misses the entire point.

The ADA interactive process was supposed to be an informal problem-solving dialogue. Instead, corporate America turned it into an immune response designed to isolate, paper over, and purge disabled workers.

Delaware’s proposed AI company form meets real shops where agents schedule staff and recommend firing. Who sets their goals, and who answers to workers?

AI agents helped me reclaim my Fire Stick. Consumers deserve control over those agents, too: accountable human review and practical local alternatives.

Cloud failures can interrupt care, travel, and ordinary work. America needs software and recovery plans that let people keep going when distant systems fail.

Noelle Tolve’s AgileRTP talk: use AI to challenge proposals, bring questions to engineers, and keep people responsible for the decisions.