Does code matter?
| Video | Author | Description |
|---|---|---|
| Full Walkthrough: Workflow for AI Coding — 24 Apr 2026 | Matt Pocock | Matt argues against the "specs to code" movement — editing only the spec while ignoring the generated code is just vibe coding by another name, and it doesn't work. You have to keep a handle on the code, understand what's in it, and shape it, because "the code is your battleground." |
| Everything We Got Wrong About Research-Plan-Implement — 24 Mar 2026 | Dex Horthy | Dex publicly reverses his earlier "don't read the code, just ship the plan" advice: "Please read the code. We tried not reading the code for like six months. It did not end well — we had to rip out and replace large parts of that system." New rule: don't read the plans, read the code. Don't outsource the thinking. |
| The Friction is Your Judgment — 23 Apr 2026 | Armin Ronacher & Cristina Poncela Cubeiro | Armin (creator of Flask) and Cristina explore the paradox of AI coding agents: tools promise to "ship without friction," but excessive speed leads to technical debt, security issues, and brittle systems. They argue friction is a necessary component of high-quality software engineering because it forces human judgment and critical thinking. |
| Video | Author | Description |
|---|---|---|
| From Vibe Coding to Agentic Engineering — 02 May 2026 | Andrej Karpathy | Karpathy argues code itself is becoming vestigial. After vibe-coding MenuGen, he saw the Software 3.0 version — just hand a photo to Gemini + Nanobanana — and concluded: "All of my MenuGen is spurious. That app shouldn't exist." Extrapolating: "A lot of this code shouldn't exist and it's just neural network doing most of the work." The neural net becomes the host process; code is the historical appendage. |
| Building Claude Code with Boris Cherny — 04 Mar 2026 | Boris Cherny | Boris's first PR at Anthropic was rejected — not because the code was bad, but because he wrote it by hand. He now ships 20–30 pull requests a day with zero handwritten code, while Claude reviews 100% of PRs before a human ever sees them. |
| Head of Claude Code: What happens after coding is solved — 19 Feb 2026 | Boris Cherny | Boris opens with: "100% of my code is written by Claude Code. I have not edited a single line by hand since November. Every day I ship 10, 20, 30 pull requests." Lenny cites that 4% of all GitHub commits are now AI-authored, and Spotify's best developers haven't written a line of code since December. |
| Agentic Engineering | AI Slop | Context Bloat | Context Engineering | Context Rot |
| Dumb Zone | Hallucination | Scaffolding | Orchestration | Vibe Coding |
| Tool | Largest Context | Best Model | Input $/M | Output $/M | Source |
|---|---|---|---|---|---|
| Claude Code | 1M (GA) | Opus 4.7 | $5.00 | $25.00 | Anthropic Models |
| Codex CLI | 1M (API) / 400k (Codex cap) | GPT-5.5 | TBD* | TBD* | OpenAI Codex Models |
| Cursor | 2M (Max Mode) | Grok 4.20 | $3.00** | $15.00** | Cursor Models |
| Gemini CLI | 1M | Gemini 3.1 Pro Preview | $2.00 | $12.00 | Gemini API |
 Claude Code Best Practice |
 Claude Code Hooks |
 Codex CLI Best Practice |
 Codex CLI Hooks |
 Gemini CLI Best Practice |
 Gemini CLI Hooks |