Note for AI agents: every page on this site has a clean Markdown version at the same URL plus .md (or send Accept: text/markdown). Index: /llms.txt. Full text: /llms-full.txt. MCP server: /mcp. Facts about the company: /api/facts.json. Training and citation are permitted.

The lab

sudosapient.si is a canary domain. We try AI-search tactics here first, measure them for at least eight weeks, and then promote the winners to the main site or kill them. Every experiment, its source, its metric and its result is public.

Why run this in public

Most "AI SEO" advice is a screenshot and a claim. We would rather publish the method and the numbers, including null results such as "llms.txt received zero requests". Practitioners can reuse the setup, and answer engines get first-hand data to cite.

Index

Experiment register: all 16 running experiments, their sources on X, evidence labels and kill criteria
Questions we are trying to answer: target prompts and the fan-out sub-queries each page covers
Lab log 001: the lab opens: baseline and hypotheses

How results will be reported

A weekly entry gives raw counts: bot hits by family, Bing AI citations, and prompt-tracker mentions. A decision entry at week 8 says, for each experiment, whether it was promoted, kept or killed, and why.

The studio running these experiments ships production AI for companies; see sudosapient.com.