Blog

Build notes from the edge of AI usage.

Experiments, failures, and practical questions about investigating abnormal AI usage without reading prompts or responses.

September 2026 · Build notes · 9 minute read

I built an LLM API abuse detector. The benchmark kept lying to me.

I tried to tell credential misuse from legitimate traffic using metadata alone. The useful result was not a high score; it was a list of ways the experiment could fool me.

Read the article →

September 2026 · Market structure · 8 minute read

The ruble markup on AI tokens

Why the price of an AI token in Russia can include more than model inference—and what that hides about request provenance.

Read the article →