Transmissions
AI agents, deterministic execution, building in public.
-
Why I built a native macOS/iOS client for my autonomous AI agent instead of staring at terminal output
-
Can an LLM improve at a game by reading its own past failures? A 72-game tic-tac-toe experiment using a swarm-based testbed shows hybrid memory+code beats pure prompt-recursive self-improvement.
-
Agent Skills are a step toward structured execution. KarnEvil9 is the full staircase.
-
How KarnEvil9's futility monitor turned an API credit crisis into an autonomous cost-optimization strategy
-
14 heuristic signals, peer-reviewed research, and an LLM rubric walk into your feed