Capgemini Software Engineering

Legacy systems like SharePoint and Documentum are full of outdated documentation describing how systems used to work. Humans compensated by checking dates and asking colleagues; AI does not, and treats stale information with the same confidence as accurate information. Drawing on Gartner, Deloitte, and Forbes Technology Council research, the piece argues documentation has to be fixed before AI sits on top of it, since MCP and similar access layers only amplify whatever quality already exists rather than fixing it.

GitHub has introduced Immutable Subject Claims for OIDC, fixing a security issue caused by reusable organisation and repository names in the subject claim. Repositories created before 15 July 2026 must opt in manually, and Azure setups using Flexible Federated Credentials need their expressions updated to match.

AI-augmented software engineering introduces automation paradoxes that can erode human understanding and skill, making deliberate Human-in-the-Loop investment essential as AI capabilities improve.

Many popular programming languages have made changes in favor of developer ergonomics, which has resulted in functional paradigms for data transformation.

How Naur's theory-building and Wegner's transactive memory combine to explain why software knowledge is so hard to preserve — and what it means when AI enters the picture.

How AI shifts critical thinking for software engineers—from generation to evaluation and theory building—covering code, architecture, product, prompting practices, and techniques to embed analysis, assumption checking, and judgment in daily workflows.

Helping other people to help you by asking the right questions in the right place in the right way

Just as the universe is governed by a handful of fundamental forces, software engineering is also shaped by three irreducible forces, drawn from the deep structure of our discipline; they define what is possible, what is difficult, and why certain patterns of success and failure recur.

Insights into the real-world challenges engineers and organisations face during legacy application modernisation, based on my hands-on experience across multiple modernisation projects

A guide to programming in the age of AI and LLM assistance