We built an iOS app using Claude Code and Codex CLI and declared a clear WINNER (no fancy workflows!) Watch now : https://t.co/n2bN6omBct

my best investment this year

my biggest realization this year is that most people making real AI products don’t talk loud about AI online

Swift Student Challenge 2026 is here https://t.co/oFb8etjjzl

funny how AI made code generation 10x easier but code quality 10x worse

LLMs love when we talk to them using DSPy

funny how AI gets paid to generate code and software engineers get paid to maintain the quality of that code

This AI coding agent is INSANELY good at UI I tried building a portfolio site using @Kombaico thinking, let's see what this has got, but the results really shocked me↓ - it first drafted a plan with specifics - it took my approval before writing code - it ran checks on the

Coding is 10% typing and 90% wrestling with your own past decisions on architecture and design patterns

Unpopular opinion : Vibe coded side projects can turn into real products if you have good sense of software engineering

i am 28 years old and i - taught dev to juniors in college at 18 - published my first YT video at 19 - made my first online $ at 20 - got first full-time job as dev rel at 23 - built a bootstrapped startup at 24 - built 30k+ developers community at 26 - help startups with dev

developers will roast me for saying this but no one cares if your app’s built with SwiftUI or React, they care if it actually makes their life easier

AI didn’t kill developers, it just proved syntax was never the hard part

funny how LLMs made code generation 10x easier but code quality 10x worse

Cursor has spoiled this whole generation of junior developers

We built an iOS app using Claude Code and Codex CLI and declared a clear WINNER (no fancy workflows!) Watch now : https://t.co/n2bN6omBct

developers will hate me for saying this, but i still prefer using CC and Amp Code over Codex CLI, because i find Codex more irritating than useful when building an iOS app

AI Agents are more flexible than humans, you can evaluate their performance, detect hallucinations and actually fix them.

The more you learn about LLMs the more you realize you'll always be required.

Chatting with LLMs feels like therapy sometimes, they just say ‘you’re absolutely right.’ No pushback, no resistance, just validation