Blog
In-depth articles on AI models, tools, and engineering practices.
Claude Web Search Tool: Dynamic Filtering, Pricing, and Implementation Guide
How Claude's web search tool works, why dynamic filtering with Opus 4.6 cuts token costs, and how to implement it in your API calls today.
Claude Code's Seven Programmable Layers: From User to System Designer
How Claude Code's seven-layer architecture turns a terminal AI assistant into a programmable coding system — context, tools, skills, hooks, subagents, and verification.
Lessons from Building Claude Code: How Anthropic Designs Agent Tools
How Anthropic's Claude Code team designs agent tools by 'seeing like an agent' — lessons on elicitation, task management, and self-directed search.
TensorFlow Still Dominates: Why Google's ML Framework Keeps Trending in 2026
TensorFlow hits 194k GitHub stars and nearly 191k commits, proving Google's open-source ML framework remains central to production AI despite rising competition.
Superpowers: Teaching AI Coding Agents to Think Before They Type
Superpowers uses structured markdown skills to force AI coding agents like Claude Code and Cursor to plan before coding — 67K GitHub stars and growing.
Claude Code Scheduled Tasks: Automate Recurring Prompts with /loop and Cron
How to use Claude Code's /loop command and cron scheduling tools to automate recurring prompts, poll deployments, and set reminders within a session.
Building a Restaurant Voice Agent with GPT-Realtime-1.5: A Practical Tutorial
How OpenAI's gpt-realtime-1.5 powers a restaurant voice agent — architecture, implementation details, and lessons learned from building production voice AI.
Red Green Refactor: Why TDD Is the Best Way to Control AI Coding Agents
Red Green Refactor — a 20-year-old TDD practice — turns out to be the most effective way to get reliable, high-quality code from AI coding agents like Claude Code.
OpenAI's Updated Model Spec: What Changes for AI Alignment and Developer Trust
OpenAI publishes an updated Model Spec defining how its models should behave — here's what changed and why it matters for developers and the AI industry.
OpenAI's Technical Lessons From Building Computer Access for Agents
OpenAI shares key engineering lessons from building computer access for agents: tighter execution loops, file system context, and secure network access.
OpenAI's Chain-of-Thought Controllability Eval: What It Measures and Why It Matters
OpenAI releases a Chain-of-Thought Controllability evaluation to measure how well reasoning models follow instructions within their thinking process.
Obsidian + Claude Code: Building a Second Brain That Actually Works
How pairing Obsidian's local markdown vault with Claude Code's file-aware agent creates a persistent, controllable context system that solves AI's biggest usability problem.