Writing.
Notes from the firm. AI infrastructure, operations, and the layers underneath.
- 01
What a Fractional Chief AI Officer Actually Does
The title sounds made up. But as of 2024, 26% of organizations had someone in this role. The role is not decorative. Here is what it actually involves.
Apr 8, 2026Christopher Swenor - 02
What Enterprises Get Wrong Before Writing a Single Line of AI Code
85% of enterprise AI projects fail to reach production. Not because the models are bad. Because nobody did the discovery work.
Mar 27, 2026Christopher Swenor - 03
Applying Karpathy's Autoresearch Pattern to RAG Retrieval
We pointed an autonomous agent at a production RAG system's knowledge base. Eight runs, $88 in agent inference cost, and the pass rate went from 22% to 89%.
Mar 23, 2026Christopher Swenor - 04
The $1.96M Routing Table: What 440 Benchmark Runs Taught Me About Picking the Right LLM
Most production AI routes every request to the same model. Usually the expensive one. 440 benchmark tests across 22 model configurations, and the per-user cost spread was 64x.
Mar 1, 2026Christopher Swenor - 05
One Line of Prompt Text Cut Our AI Costs by 86%
Two days benchmarking 22 model configurations across 440 runs. Then five minutes changing one line of prompt text. That edit cut per-user costs by 86%.
Mar 1, 2026Christopher Swenor - 06
We Tested Reasoning Mode Across 160 Runs. It Improved Nothing.
Reasoning mode tested across 8 models and 160 dedicated benchmark runs. It improved zero of our four workflows. One provider silently disables it when tools are present.
Mar 1, 2026Christopher Swenor - 07
Why AI Costs Explode at Scale and the Architecture That Controls Them
A prototype that cost $14 over a weekend would cost $40,000 a month at scale. The same system can cost $4,000 with the right engineering patterns.
Mar 1, 2026Christopher Swenor