Essays and field notes on AI, software engineering, design, and the craft of building product teams that ship. Written by the engineers doing the work.
MIT Project NANDA found vendor-built enterprise AI pilots succeed 67% versus 33% for internal builds, while IDC and Deloitte pegged failure at 88% and 89%.
Cursor 3 spawns up to 8 AI agents in isolated git worktrees, while Anthropic's Claude Opus 4.8 dynamic workflows lift the per-run subagent ceiling to 1,000.
Surge AI bootstrapped to $1.2 billion in 2024 revenue, surpassing Scale AI's $870 million by charging up to 10x for frontier-lab RLHF work.
LangChain's LangGraph orchestration layer now runs in 400 production enterprises including Cisco, Uber, LinkedIn, BlackRock, and JPMorgan.
Google made Gemini 2.5 Pro Deep Think generally available on June 22, 2026, with a 2-million-token context window and 76.4% on SWE-bench Verified.