Why Coding Agents Are Moving to the Terminal
CLI-based coding agents are overtaking IDE tools because they need access to the real development environment—databases, tests, git, and shell commands—not better code display.
★ YOUR NEIGHBOUR UNDERSTANDS AI. DO YOU? ★
The Journey

Study
Flashcard decks. Five minutes a day. Six knowledge areas.
Build
Step by step guides. Real projects. No CS degree required.
Update
Short reads on what matters in AI. No hype, no jargon.
Flashcard decks. Five minutes a day. Six knowledge areas.
Each deck covers a knowledge area that matters for working with AI. The cards are short, the sessions are five minutes, and the spaced repetition engine makes sure things stick.
You do not need to study everything at once. Pick a domain, open the app, and start where you are.

Available on iOS
Step by step guides. Real projects. No CS degree required.
Each guide walks through a real project, step by step. You will build something that works, not just read about how it could work.

A complete walkthrough of building FlowDoc from empty repo to signed desktop app using AI pair programming.
Start learning
Build sophisticated AI workflows using advanced prompting techniques like Chain of Thought, ReAct, and Super-Prompts to create context-aware AI assistants that understand your specific needs and domain expertise.
Start learning
Switch from Windows to Ubuntu and set up a development environment optimized for AI work. Learn why Linux matters for AI, install Ubuntu safely, and get your tools running.
Start learningShort reads on what matters in AI. No hype, no jargon.
Short reads on what actually matters. Each insight covers one idea, explains why it's relevant, and skips the breathless hype cycle.
CLI-based coding agents are overtaking IDE tools because they need access to the real development environment—databases, tests, git, and shell commands—not better code display.
Top engineers at Anthropic and OpenAI say they've completely stopped writing code. The shift reveals something uncomfortable about what programming actually was.
OpenAI's latest model tops every benchmark but can't tell when it's wrong. The gap between what AI can do and what it should do is widening.