AI has made software more expensive, not cheaper, because companies reinvest productivity gains into new features and higher tiers rather than passing savings to customers, according to an essay by an…
AI has made execution nearly free, shifting scarcity from engineering labor to decision-making, and the next frontier is wanting—the goals that drive decisions. As AI improves at deciding, humanity's …
A new divide in software interfaces is emerging between chat-first and UI-first users, but the real distinction is whether the user or the software holds the task model. Chat interfaces offer freedom …
The rise of AI agents is shifting the primary work device from laptops to phones, as human roles narrow to decision-making and approval rather than high-bandwidth output. The bottleneck moves from typ…
AI review and evals serve different purposes: review inspects the change (diff), while evals inspect the finished artifact in operation. When AI regenerates entire pages, the diff becomes unmanageably…
Production agentic AI deployments rely on robust harnesses—scaffolding for timeout, retry, observability, and rollback—rather than the model itself. The key research question for 2026 is finding minim…