How it works Guides Journal Pricing

Open Infinite Graph

How it works Guides Journal PricingFree · Premium $12.90/mo

JOURNAL / NOTES ON KNOWLEDGE WORK

Ideas we test
before we build them.

Essays on personal knowledge management, AI memory, and knowledge graphs — written while building Infinite Graph.

September 24, 2026Graph Design Is Identity Design Before It Is Relation DesignBefore you build a knowledge graph, you have to answer a surprisingly philosophical question: what counts as one thing?↗︎September 24, 2026PKM Stores Nouns Well. It Stores Surprisingly Poorly.Most knowledge tools are good at storing what things are. They are much worse at storing what happens.↗︎September 24, 2026AI Editing a Knowledge Graph Needs Transactions Before It Needs IntelligenceAn AI editing a knowledge graph without transactions is basically a database that can stop halfway through `COMMIT`.↗︎September 19, 2026A Knowledge Base Should Remember What You Decided Was WrongA knowledge base that forgets rejected ideas will eventually make you rediscover the same bad ideas.↗︎September 15, 2026Uncertainty Should Have Dependencies TooOne uncertain assumption can quietly support thirty confident conclusions.↗︎September 13, 2026The Better the Summary, the Easier It Is to Lose the SourceA perfect summary can make the original document effectively disappear.↗︎September 13, 2026The Architecture of Software Is Not the Architecture of UnderstandingThe architecture of a codebase and the architecture you need to understand it are not necessarily the same graph.↗︎September 13, 2026A Mature Knowledge System Needs Garbage CollectionMaybe the hardest feature in a knowledge system isn’t capture. It’s deletion.↗︎September 11, 2026The Real Reason It Is So Hard to Switch PKM ToolsThe real lock-in in a knowledge tool is not your files. It is the hundreds of tiny habits you have built around it.↗︎September 8, 2026The Most Important Part of a Meeting May Be What Wasn’t DecidedMeeting notes record what was said. They rarely record what was deliberately left undecided.↗︎September 7, 2026What If a Knowledge Graph Should Remember That Two Things Are Not Connected?Knowledge graphs store relationships. They rarely store relationships that were explicitly rejected.↗︎September 7, 2026The More Dangerous AI Error: When Explanation Becomes RationalizationA wrong AI answer is sometimes obvious. A plausible explanation for a wrong answer is much harder to notice.↗︎August 30, 2026From storage and retrieval to timely reappearanceI don’t think I have a note-taking problem anymore. I have a resurfacing problem.↗︎August 29, 2026Read access and write authority are being treated as the same AI featureRead access and write authority are being treated as the same AI feature. They shouldn’t be.↗︎August 28, 2026A good summary and a good knowledge structureA good summary and a good knowledge structure are almost opposite optimization problems.↗︎August 28, 2026You Saved It Somewhere. Good Luck Finding It Again.We live in an age in which saving information has become almost trivially easy.↗︎August 27, 2026Doesn’t AI memory actually need a shelf life?AI memory has a deletion problem, but it also has a staleness problem.↗︎August 24, 2026Your Knowledge Graph Is Beautiful. It Is Also Probably Useless.Why I stopped treating the graph as a visualization and started building it as an interface for thought.↗︎

Infinite GraphA knowledge graph that stays yours.Free · Premium $12.90/mo

LegalPrivacy Policy Terms of Service Refund Policy

Supportcontact@infinitegraph.app

MoreOpen Infinite Graph Journal GitHub© 2026 Infinite Graph