Interactive Explainers

Clear, interactive essays on topics in AI, science, and computing. You don't really understand something until you can play with it.

• Embedding OpenAI Codex: The App Server and SDKs

  Codex is not just a CLI. Under every surface sits an app-server speaking JSON-RPC, plus two programmatic doors into it: the protocol itself and the SDKs built on top.

  Interactive July 2026 17 min read 5 figures

• Stagehand: Inside the AI-Driven Browser Automation Framework

  Hardcoded selectors break on every redesign; autonomous agents are unpredictable. Explore how Browserbase's Stagehand splits browser automation into four AI-resolved verbs - and caches their answers so you only pay for intelligence when the page changes.

  Interactive July 2026 10 min read 5 figures

• Superpowers: The Anatomy of an Agent Skill

  AI agents will skip your guardrails the moment they feel inconvenient. Explore how Superpowers - the 200k-star agent-skills framework - bootstraps itself into every session and what makes one skill stick where another gets quietly ignored.

  Interactive May 2026 13 min read 4 figures

• Artemis II: Why Going Back to the Moon Is a Big Deal

  On April 10, 2026, four astronauts splashed down after the first crewed lunar flyby in over 50 years. Apollo already did this in 1968. Explore what Artemis II actually proved, how Orion works, and why this mission reshapes the next decade of human spaceflight.

  Interactive April 2026 14 min read 5 figures

• DSPy: Programming - Not Prompting - Language Models

  Stop hand-tuning prompt strings. DSPy lets you declare what your LM should do, compose calls into pipelines, and compile optimized prompts automatically. Explore signatures, modules, and optimizers interactively.

  Interactive April 2026 15 min read 3 figures

• Autoresearch: AI That Does ML Research While You Sleep

  Give an AI agent a real LLM training setup, go to sleep, wake up to a better model. Explore how Karpathy's autoresearch lets agents run 100 experiments overnight - keeping wins, discarding losses, and steadily improving.

  Interactive March 2026 15 min read 3 figures

• Pi & OpenClaw: The Self-Extending Agent

  Most coding agents ship with dozens of tools. Pi ships with four - and asks the AI to build the rest. Explore how this minimal agent architecture powers OpenClaw's multi-channel AI platform.

  Interactive March 2026 12 min read 4 figures

• World Models: How AI Learns to Simulate Reality

  Language models predict the next word. World models predict the next moment. Explore how Google's Genie, LeCun's JEPA, and Fei-Fei Li's World Labs are building AI that understands physical reality.

  Interactive March 2026 15 min read 5 figures