Link Dump #232

Did you already read something interesting this week? If not, here's an opportunity to change it.

1. Software Architecture
   1. Software Architecture must evolve - or it becomes the Problem #PickOfTheWeek
      Architecture is often treated as something written in stone. In reality, software continuously changes - and architecture must evolve with it. The real danger isn’t changing architectural decisions. The danger is when they never change at all.
2. Software Development
   1. Coding Is Solved
      The purpose of this page is to document bold public predictions and let time be the judge. It's a showcase of how marketing and selling dreams works in the tech industry - where hype cycles often outpace reality. Accountability matters, especially when predictions shape hiring decisions, career choices, and public policy.
   2. AI Doesn't Fix Your Real Bottleneck #PickOfTheWeek
      Why are we using AI to solve the easiest part of the job? The author breaks down why 'coding' is merely the mechanical translation of a design into a language a computer understands. If your design is flawed or your domain boundaries are blurry, AI will simply help you build the wrong thing faster.
   3. Spec-Driven Developmentwith AI Coding Agents: The Workflow Replacing”Prompt and Pray”
      Spec-Driven Developmentwith is the discipline of making your architectural standards (transaction management, exception hierarchies, record types) explicit in a Markdown file. Learn how to build a 'requirements.md' that prevents hallucinations and ensures your AI-generated microservices actually look like they were written by your team.
   4. Three Amigos with AI: Stop building the wrong thing faster #PickOfTheWeek
      Justin Searls explains why the Three Amigos - Product, Development, and Testing - must now act as the 'Prompt Engineering' layer of the business. Learn how to stop using AI to build features faster and start using it to build the right features with 100% alignment. 
   5. Context Anchoring
      Why does AI give generic advice when you need specific solutions? The problem isn't the AI's intelligence; it's a lack of Context Anchoring. Learn how to stop treating the LLM like a magic box and start treating it like a researcher that needs a library. Discover how to 'anchor' your prompts with your actual codebase, architectural ADRs, and domain models to get better results. 
3. Agile 
   1. Product Backlog Refinement: How Scrum Teams Do It Right #PickOfTheWeek
      Is your refinement session just a long, boring meeting where everyone stares at Jira? The author breaks down why the most successful teams treat refinement as a continuous flow. Learn how to move away from the 'marathon session' and toward a 'just-in-time' model that keeps the backlog healthy without draining your team's energy. 
4. Growth
   1. This has happened before. It's happening again.
      Historically, every major tech shift has created more jobs for developers. This time is different. Doc Norton explains why Agentic Coding - AI that reasons, refactors, and tests - is a compression event. One developer with agents can now do the work of an entire team.