Reading Notes
Links, short summaries, and notes on what is worth noticing across AI, operations, strategy, and company-building.
A framework for the next decade of software: moving from writing code to building loops, harnesses, and factories. The unit of software work is moving from the code diff to the agent loop and its harness.
Shows how Cursor uses multi-agent swarms to handle complex coding tasks (like building SQLite from scratch) by separating planning from execution. Cursor grouped models into hierarchical swarms to handle larger, more complex tasks.
Stronger sales execution starts by cutting distractions and concentrating on the few priorities that drive revenue. Build a system around that focus before adding AI or increasing activity.
Chinese open-weight models are unexpectedly competitive. This changes the balance of power in global AI. Chinese startup Moonshot AI released Kimi K3, a 2.
How to build AI-native teams for speed instead of headcount. Successful AI-first companies are dropping traditional departments for "mission pods."
Replit's CEO explains how integrating agents into the company tripled engineering code output while maintaining quality. Over the last six months, Replit embedded agents into its workflows.
Early proof that AI agents can iterate on their own code, suggesting capabilities will compound faster than expected. Researchers built AIDE², an AI that improves itself through two loops: an inner loop for code and an outer loop for the agent's harness.
AI engineering is shifting from standalone autonomous agents toward the infrastructure around them: workflows, context, evaluations, permissions, and state. The practical goal is reliable systems that augment human operators, not autonomy for its own sake.