All news on this site, in one place.
News

Engineers must retain full accountability for AI-assisted code because models cannot know system-specific failure modes.

Platform engineering creates a shared layer to manage the identity, budget and audit trails of LLM applications.

The software factory architecture manages the increase in agent-led code production by focusing on the delivery system.

AI agents can automate pull request merges, reducing the technical necessity of the two-week sprint box.

DeepSeek and Huawei are releasing open-source programming tools for Ascend chips to lower the barrier for developers.

Zero-shot forecasting models provide a superior alternative to traditional tuning by prioritising reuse over maintenance.

Vetting a development partner involves assessing the reasoning process rather than just the initial proposal.

Measuring AI productivity through volume alone ignores the costs of rework and review queues.

Governance for AI agents must move from static documents to integrated controls within the delivery pipeline.

Software craftsmanship treats development as a skilled trade based on technical mastery and ethical responsibility.

Verification capacity has failed to keep pace with the surge in code generation capacity provided by AI coding agents.

PHP 8.6 introduces partial function application and a new Duration class to reduce boilerplate and improve precision.

Software delivery constraints have shifted from the act of writing code to the process of proving that code is safe to release.
