Case Study: Reviving a 2016 Android App with an AI AgentEvery developer eventually inherits a codebase they didn't write, can't fully explain, and are expected to ship changes to anyway. Add an AI coding agent into that mix and you get two very different oSep 10 · 5 min readmarkdownsimplified.hashnode.dev
Beyond Prompts: How Agentic AI Actually Works (Especially for Coding)A developer once handed an AI agent a single-line instruction — "fix the failing tests" — and walked away for coffee. Ten minutes later, the agent had rewritten half the test suite, "fixed" it by deleAug 7 · 5 min readmarkdownsimplified.hashnode.dev
The Real Cost Table: 10 Prompt Patterns That Are Secretly ExpensiveMost people blame the model when their AI bill creeps up. It's rarely the model. It's usually one of a small set of habits — vague scope, re-explaining things the model already has, correcting in piecAug 1 · 4 min readmarkdownsimplified.hashnode.dev
Two Prompting Modes: Explorer vs. ExecutorA few weeks ago I watched two developers use the same AI agent, on the same day, with completely different results. One spent twenty minutes going back and forth — "no, not like that," "actually can yJul 27 · 4 min readmarkdownsimplified.hashnode.dev
Why Your AI Costs More Than It ShouldTwo developers ask an AI agent to do the exact same task. One spends four back-and-forth messages fixing small misunderstandings. The other writes one detailed prompt upfront. Same outcome, same modelJul 24 · 4 min readmarkdownsimplified.hashnode.dev
