We publish field notes about AI discovery, agent systems, knowledge environments, and computational craft. The aim is to connect an idea to a choice someone can make: what to build, what to test, what to read, and what to leave alone.
The work is written for builders, creative studios, and teams making useful systems visible. A shared interest is a starting point, not a claim that every reader has the same needs.
The editorial method
Every note should answer a concrete question, add a usable contribution, and link to the sources behind its factual claims. Recommendations and hypotheses are distinguished from results. We update existing notes when the question stays the same.
Research and drafting use AI assistance. The /init editorial byline identifies the publication, not a fictional person or an assertion of human review. Source checks establish what a document supports; they are not independent reproductions of a paper's results.
What earns publication
- A meaningful problem and a clear audience.
- Public sources checked for the claims they support.
- An explanation, method, comparison, or experiment that adds value.
- Accurate dates, working links, and readable output on mobile.
Corrections and revisions
Substantive updates change the visible revision date. Evidence gaps remain visible. We do not invent customer outcomes, imply that an unrun experiment succeeded, or promise rankings or AI citations. Private working notes and operational records are excluded from the site build.
A connection to the work
Explore Metanomicon ↗
Read FORM//GRAPH ↗
Follow without an account
Use the RSS feed. The JSON index and Markdown versions contain the same public articles. This site sets no application cookies and includes no third-party analytics scripts; the hosting provider may process request logs to serve and secure it.