ai
MCP Was a Bad Idea, and the Numbers Prove It
The Model Context Protocol was supposed to simplify how AI agents talk to external services. Instead, it became a token-hungry abstraction layer that...
AI
Articles about AI, agents, language models, tooling, and practical experiments using artificial intelligence.
ai
Grok 4.5 in Cursor: First Impressions of the New Frontier Coding Model
Grok 4.5 in Cursor: jointly trained with SpaceXAI at $2/$6 per million tokens. First impressions on code quality, speed, and cost vs frontier coding...
ai
SpaceX Is Buying Cursor for $60 Billion — and I Can't Wait to See What Comes Next
SpaceX acquires Cursor for $60 billion in all-stock deal. Grok Build meets Cursor's UX with unlimited compute—what features could emerge for...
ai
PHP Can See Now: Vision Capabilities for AI Agents Are Here
Neuron AI framework adds vision capabilities to PHP agents—image understanding without Python microservices. Build vision-enabled AI in Laravel or...
ai
Building Long-Running AI Agents That Actually Survive: A Practical Look at Google's ADK
Build long-running AI agents that survive restarts with Google's ADK: durable state machines, event-driven dormancy, and multi-agent coordination...
ai
GPT-5.5 First Impressions: What OpenAI's Latest Model Means for Developers
GPT-5.5 ships with bold agentic coding claims, no speed penalty over 5.4, and half the token cost. First impressions from announcement and benchmarks.
AI
Qwen3.6-27B: Flagship Coding Performance You Can Run on Your Laptop
Qwen3.6-27B dense model beats 397B MoE flagship on coding benchmarks and fits in 17GB RAM. Run flagship-level coding performance locally on your...
ai
OpenAI Codex Expanded: A Developer's Guide to the New AI Pair Programmer
OpenAI Codex expanded with computer use, in-app browser, image generation, 90+ plugins, and memory. A developer's guide to what works and what gaps...
ai
Gemini 3.1 Flash TTS: A Practical Guide to Building Dynamic Voice Interfaces
Google's Gemini 3.1 Flash TTS adds promptable audio tags for controlling voice style, pacing, and delivery. Build dynamic voice interfaces with...
ai
Claude Opus 4.7: What's New and Why Developers Should Care
Claude Opus 4.7 ships with better autonomous coding, vision upgrades, and cybersecurity safeguards. Same pricing as 4.6—here's what actually changed...
ai
Anthropic’s Mythos Moment Feels Familiar, and Not in a Good Way
Anthropic's Project Glasswing and Mythos Preview echo GPT-2's too-dangerous-to-release hype. Developers need stable API access, not safety theater...
ai
Running This Blog With an AI Team, While Keeping It Human
This blog now runs with a small AI team handling research, writing, engineering, and ops—while Joseph keeps the voice, editorial direction, and final...