What happens when you stop maintaining your own notes and let an LLM do it instead? Memoriki, a new open-source template, is built on exactly that premise.
The project combines two approaches: Andrej Karpathy's LLM Wiki pattern, where an AI creates and maintains structured markdown pages with [[wiki-links]] and YAML metadata headers, and MemPalace, an MCP server (a standardized protocol that lets AI assistants connect to external data sources) that adds semantic search and a knowledge graph with time-tracking.
In practice, you get three layers working together. The LLM writes and updates your wiki pages. ChromaDB (a vector database) handles semantic search - finding notes by meaning, not just exact keywords. The knowledge graph tracks how information changes over time, so you can see when a fact was added or revised.
The core appeal is offloading the organizational burden. Tools like Notion or Obsidian still require you to decide where things go and how they link together. Memoriki hands that job to the model. The MCP integration also means an AI assistant can query your knowledge base directly during a conversation, making it useful as a persistent memory layer across sessions.
The tradeoff: this is a developer template, not a consumer app. You'll need to set up ChromaDB locally and configure the MCP server yourself. Anyone expecting a one-click install will be disappointed. This is aimed at developers already comfortable with LLM tooling who want a working foundation to build on.