Software-engineering

Shopify's September 2026 return to Swift and Kotlin is not a React Native failure. AI is changing the cost of building mobile apps twice, which will split the market: native at the high end, Expo for everyone else, and brownfield in between. The most valuable mobile engineers will be systems people who can orchestrate agents across Swift, Kotlin, and React Native.

In 2026, elite engineers are not measured by lines of code or hours worked. They deliver value through speed, quality, and AI leverage — shipping small changes, removing waiting, and focusing on outcomes over activity. This post explores the new formula for engineering excellence and the habits that separate average teams from elite ones.

For more than 20 years, Agile dominated software delivery. AI has changed the cost, speed, and structure of building software — yet most companies still run ceremonies designed for slow, expensive, communication-heavy coding. This post argues that Agile is becoming operational theater: story points collapse, velocity becomes vanity, and the bottleneck shifts from implementation to approvals and process. The future is outcome-driven, post-Agile engineering — continuous delivery, AI-assisted execution, tiny PRs, and throughput over ceremony.

Today's pull request is built around a diff. You review what changed, not how it came to be. With coding agents, that model is starting to leak: a clean diff can hide a messy session, skipped constraints, or a plan that was wrong for three iterations before it was right. This post explores a new kind of PR — one where the artifact under review is the agent session itself: the prompts, the tool calls, the rejected paths, the plan, and the human's choices along the way. It compares the idea to existing practices like AI review bots, stacked diffs, build provenance, and "session provenance", and argues that making the session the primary review surface is genuinely new — and probably inevitable.

A rolling manifesto for building software with AI agents, in the spirit of the 2001 Agile Manifesto.

AI-generated outputs from tools like Claude and GitHub Copilot are not independent artifacts but the direct result of how they are guided through prompts, context, and constraints. This means the engineer fully owns both the strengths and flaws of the output. Selective attribution, where success is claimed and failure is blamed on the model, is inconsistent and undermines standards. Effective use of AI requires deliberate input, rigorous review, and full accountability, with the understanding that anything produced and shipped is ultimately the engineer’s responsibility.