Karpathy's LLM Wiki Pattern, Hosted

Your research, compiled into a living knowledge base

Drop in URLs, PDFs, and notes. Claude compiles them into a structured, interlinked wiki you can chat with. No vector DBs. No RAG pipelines. Just knowledge that compounds.

Free tier: 1 wiki · 20 sources · 100 queries/month

How it works

Unlike RAG which re-reads raw docs on every query, WikiMind builds compiled knowledge that gets better with every source you add.

01

Drop in Sources

Paste URLs, upload PDFs, or type notes. WikiMind fetches clean content from any public URL.

02

Claude Compiles

Claude reads your sources and writes structured wiki pages with interlinks, tags, and summaries.

03

Browse Your Wiki

Explore a growing graph of interconnected pages. Each source enriches the whole knowledge base.

04

Ask Questions

Chat with your compiled wiki. Claude reads pages (not raw docs) to give precise, cited answers.

Raw SourcesCompileWiki PagesQueryAnswer

Knowledge compounds over time — each new source makes every answer better.

Simple pricing

Start free. Upgrade when your wiki outgrows it.

Free

$0

Perfect for a single research project.

  • 1 wiki
  • 20 sources
  • 100 queries/month
  • URL & text ingestion
  • Full wiki browsing

Pro

Popular
$19/mo

For serious researchers and knowledge workers.

  • 5 wikis
  • 500 sources per wiki
  • Unlimited queries
  • PDF upload
  • Wiki lint & health check
  • Priority AI compilation