My favorites
Agile, Leadership and Product
- Skip the ChatGPT-for-contracts trap: purpose-built legal AI beats public models on security and clause accuracy. Worth a skim before your legal team picks a tool.
- Behavioral interview questions beat ‘tell me about yourself’ every time—here’s a free, open-source bank of 200+ to steal for your next hiring loop.
- Annual engagement surveys are dead — pulse checks tied to real moments and managers trained to listen beat once-a-year scores every time.
- ThoughtWorks’ latest read on where European software engineering is headed — worth a skim if you’re setting team strategy for the next few years.
- Cagan points to Benedict Evans’ take on why most people won’t vibe-code their own tools — a sharp reframe of what product managers actually contribute in the AI era.
- The real trick with vague asks: stop hunting for solutions and spend your energy nailing the problem definition first — clarity beats speed.
- Great reframe for staff+ engineers: stop scheduling ‘strategic thinking’ time and start absorbing the problems people complain about daily — the good ones surface themselves.
- Sobering read if you’re eyeing the exec track: middle management often trains the wrong muscles, and too much time there can actually hurt your shot at it.
- Retention isn’t about counteroffers — this breaks down the real barriers that keep your best people from walking, before they’re already out the door.
Architecture, Development & Software development practices
- A deep dive into pushing React Server Components further than the usual demo — URL-driven pagination, streaming search, and server-rendered previews without client-side fetching.
- A methodical, code-first walkthrough of concurrency models for network servers — great primer before you pick threads, async, or event loops for your next service.
AI, LLM & Machine Learning
- Zalando’s 2.5-year playbook for scaling LLM adoption: LiteLLM proxy for multi-provider access, cost tracking hooks, and a CLI born from a hackathon.
- AI ideation tools don’t fix bad brainstorming, they amplify it. Worth reading before you assume more AI will make your team more innovative.
- A genuinely different take on RAG: skip the vector DB and let an LLM reason over a document’s actual structure like a human flipping to the right section. Worth a look if similarity search keeps missing what you need.
DevOps, Observability & Security
Tools and things from Github
- A deep dive into building a VS Code bookmark extension the right way — tracking points over line numbers, cached tag snapshots, and other lessons for anyone building editor extensions.
- A CLI that scans your RAM, CPU, and GPU, then tells you which of hundreds of local LLMs will actually run well — no more guess-and-crash downloads.
- A native task manager that goes deeper than Activity Monitor or Task Manager—per-core, thermals, energy, and process trees, built once and shared across macOS, Windows, and Linux.