Chapter II The Book of Life
AI & Automation
Working with AI day to day: coding assistants, vibe coding, LLM graders, context windows, and the jobs AI is changing.
This chapter is about using AI in real work, not about AI in the abstract. Most of it comes from a year of building with coding assistants and agents, and from measuring where they help and where they quietly make things worse.
11 essays in this chapter
- How AI Agents Changed My Development Workflow: From 13 Years of Coding to Orchestration I started with Udemy courses and StackOverflow panic. Now I run five AI agents across five projects from a CLI. Somewhere in between, my entire thought process changed.
- LLM-as-a-Judge Reliability: Check Your Grader Against Human Labels LLM-as-a-judge reliability depends on more than an 80% agreement score. Here's what to check before you trust an AI grader.
- When an Upstream API Changes, Your Agent's Tools Break Silently API breaking changes detection fails for AI agents because the break returns 200 OK. Why versioning and contract tests miss it, and what catches it.
- AI-Assisted Coding vs Vibe Coding: The Difference Is Review Vibe coding and AI-assisted coding use the same tools and produce very different software. The line is review, and the data on what happens when you skip it.
- AI Context Window Too Small? Why Long Prompts Lose Accuracy Your AI context window feels too small even at 200K. Lost in the Middle found 20-30% accuracy drops when the answer sits in the middle of a long prompt.
- Agent Rule Conflicts: Why AI Pipelines Fail Silently When AI agents have conflicting rules, pipelines fail silently. No exception, no stack trace. How cascading deferrals break automation, and what to audit.
- New Job Titles AI Is Creating, Not Just Destroying AI is killing jobs and creating stranger ones. Six new AI job titles paying real salaries in 2026, what each one does, and which will last.
- Is AI Pair Programming Worth It? One Year With Copilot, Cursor and Claude Is AI pair programming worth it? After a year with GitHub Copilot, Cursor and Claude: the speed gains are real but concentrated, the bug data is sobering, and authorship quietly moves upstream.
- Vibe Coding vs AI-Assisted Development: What Is the Difference? Vibe coding vs AI-assisted development: same tools, opposite workflow. The difference is whether you read, understand, and test the code before you ship.
- Silent AI Code Bugs: Passing Reviews, Failing in Production AI-generated code looks right and still ships bugs review misses. Automation bias is why. How to audit it before three weeks of silence becomes an incident.
- What AI Agents Still Cannot Replace About Senior Developers AI agents write code fast, but they can't replace the judgment, systems thinking, and context that senior developers bring to real problems.