Writings
Thinking out loud about AI that lives in the real world
Notes, essays, and field reports from building AI characters and experiences that survive contact with reality. From quick thoughts to deep dives on narrative systems, operations, and the messy intersection of creativity and code.
Filter:
technical
ai-performance
ai-art
KWLX Deep Dive: Actor/Character Framing
•
15 min
Why Your AI Characters Turn to Mush (And How I Fixed It)
tools
open-source
ai-ops
Introducing Qualspec! (Merry Christmas)
•
10 min
Introducing Qualspec: LLM-Judged Testing for Ruby
tools
open-source
ai-ops
Open Router Enhanced: Updated to 1.2
•
10 min
OpenRouter Enhanced 1.2: What’s New (and Why It Matters for Character Work)
ai-safety
mental-health
rlhf
New Essay: Role-Based Reality
How AI withholds life-or-death information unless you know the magic words. A modest experiment exposing the paternalism at the heart of RLHF-based AI safety.
•
2 min
I just published a new piece on my Substack exploring something that’s been bothering me for a while: the way AI systems gate critical information behind role claims.
tools
open-source
ai-ops
Duct Tape DSLs: Model Routing, Structured Outputs, and Typed Tools
Deploy a real-time character and you learn quickly: everything breaks, always, and in ways that don't exist in web-land.
•
10 min
3 a.m. on the KWLX stream: Figment’s mid-sentence about medieval polyphony when the model I was using goes down. The pipe goes silent. Dead air. The magic dies in 3000ms.