Writing
What fits, what it costs, and why it is slow
Local LLMs, GPU sizing, GGUF quants, and what AI agents cost. The local-AI numbers come from dated runs on hardware I own. Every other figure names its source.
- 143 pull requests. Zero dollars.6 min
- My Local Blog Writer Drops Private Lines First5 min
- Ollama JSON: Empty Results Are Not Failed Requests5 min
- A kill rule that expires is not a kill rule6 min
- Why Your Local Model Fits and Still Fails at Long Context5 min
- Prove llama.cpp Tensor Split Used Every GPU5 min
Start here
The questions people ask most, answered.
real test outputMy 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
143 pull requests. Zero dollars.
In the week of 2026-08-31 the fleet merged 143 product pull requests. Stripe still read zero. Throughput is the tell, not the win.
- 5 min read
My Local Blog Writer Drops Private Lines First
My local blog writer reads a daily log. I remove private lines before drafting, then check the output again. Here is what those checks can miss.
- 5 min read
Ollama JSON: Empty Results Are Not Failed Requests
An empty Ollama result can mean no matches or a broken request. I test the response parser so local extraction failures cannot pass as clean results.
- 6 min read
A kill rule that expires is not a kill rule
I wrote a kill rule for a paid-path test. Zero orders landed. The card expired into an archive and the dashboard still said ACTIVE.
- 5 min read
Why Your Local Model Fits and Still Fails at Long Context
A local model can load and still run out of memory at longer context. Compare two controlled loads, inspect cache logs, and test the real workload.
- 5 min read
Prove llama.cpp Tensor Split Used Every GPU
A tensor-split flag is only a request. Pin the split, watch every GPU, and save one repeatable llama.cpp receipt before trusting the result.
- 5 min read
Local or API? Test the task before routing it
Compare a local model with an API on the same task. Record settings, failed attempts, review time and cost before you decide where the workload belongs.
- 5 min read
Log Local LLM Fallbacks Before You Score the Output
My local writer hit a host RAM guard, then Sonnet produced a draft that failed QA. Here is how I separate availability, model identity, and useful output.
- 5 min read
Agent Memory: Test the Answer After a Correction
Our agent-memory replay checks words in a file. I show what that proves, what it misses, and how to test whether an expert correction changes an answer.
- 5 min read
My local daily brief used 0 model calls
On September 4, 2026 my llama3.1:8b daily brief was ready with 0 model calls and 0 seconds. Preview-only file ranking is not a model run.
- 6 min read
My agent wrote 126 empty pages and every gate passed
One commit wrote 130 knowledge pages. 126 were the same four sentences with the title swapped. Schema checks, link checks and orphan checks all passed.
- 5 min read
Long Context Cost Me VRAM, Not Tokens a Second
A 94x bigger prompt cost llama3.1:8b only 9.7% of its generation rate. VRAM rose 656 MiB, but num_ctx doubled in the same step. Budget memory first.
The AI agent build notes
Real costs, real tools, no fluff. One evidence-backed note on Friday when there is something worth sharing.
Get the requested artifact now, then at most one evidence-backed Local AI Lab Note on Friday when there is something worth sharing. One-click unsubscribe. No sponsored placements. Privacy.