Software Engineering & Architecture At the AI Dev 26 x SF conference last week, speakers argued that AI has fundamentally inverted the development bottleneck: writing code is no longer the constraint—imagination and clear specification are. The Register's coverage highlighted a growing consensus around spec-driven development, with practitioners finding that giving models well-formed specifications dramatically improves output quality. A secondary theme: reducing agent defect rates matters more right now than pushing capability frontiers, since deployment at scale amplifies every error. The future of software development has less development
Global Macro & Markets Oil surged ~6% today to $114.44/barrel (Brent) after Iran escalated attacks on UAE oil infrastructure and struck vessels in the Strait of Hormuz—the most serious flare-up since the early-April ceasefire. Trump's "Project Freedom" naval escort initiative launched yesterday but has moved only four ships through the strait, against a pre-war average of 120 per day. The spike complicates the Fed's rate calculus: the FOMC held steady at 3.5–3.75% at its April 29 meeting in an 8–4 split, with the minority fearing tariff-plus-energy inflation could re-accelerate. Oil prices surge as violence flares in Strait of Hormuz
Enterprise AI & AI Strategy IBM opened Think 2026 today with a pointed framing: the "AI divide" between leaders and laggards is widening, and the gap is no longer about access to models—it's about operating model design. The flagship announcement is a next-generation watsonx Orchestrate in private preview, repositioned as an "agentic control plane" that can deploy and govern agents from any vendor under a unified policy layer. IBM's argument is that enterprises need infrastructure to manage heterogeneous agent ecosystems, not just more capable individual agents. Think 2026: IBM Delivers the Blueprint for the AI Operating Model
AI Tools & Developer Utilities Anthropic quietly announced Claude Mythos Preview, a new model with unusually strong performance on computer security tasks, alongside Project Glasswing—an initiative to use Mythos to help harden critical open-source software. The model is being rolled out to a limited set of critical infrastructure partners and open-source maintainers first. Separately, Anthropic released a broad Claude Code update covering smarter model selection, project purge tooling, stronger permission handling, and a long list of stability fixes across Windows and PowerShell environments. Claude Mythos Preview and Project Glasswing