---
mw_bundle: 1
id: ggAzbcHr
title: "memory.wiki: Knowledge Graph as AI-Native Infrastructure"
url: https://memory.wiki/b/ggAzbcHr
document_count: 3
updated: 2026-05-21T02:50:47.440Z
analysis_generated_at: 2026-05-21T02:50:47.440Z
source: "memory.wiki"
---
# memory.wiki: Knowledge Graph as AI-Native Infrastructure

> The vision, architecture, and business rationale behind memory.wiki — a personal knowledge wiki that serves as a URL-based memory layer for any AI.

## Summary

These documents chronicle the comprehensive development of memory.wiki, a personal knowledge wiki that functions as a deployable memory layer for AI systems. The collection traces the evolution from the initial mdfy concept through multiple business plan iterations to a finalized product specification, consistently centered on the principle that knowledge should be URL-accessible to any AI platform.

## Themes

- URL-native knowledge delivery
- Cross-AI platform strategy
- User authoring vs AI extraction
- Knowledge graph automation

## Cross-document insights

- The reframing from 'AI memory problem' to 'knowledge delivery problem' represents a fundamental strategic insight that differentiates memory.wiki from competitors like Mem0 and Letta
- The structural impossibility for big tech companies (OpenAI, Anthropic) to build cross-AI solutions creates a sustainable competitive moat for independent developers
- The tension between comprehensive vision and shipping constraints led to multiple scope revisions, revealing the challenge of balancing ambition with execution timelines
- The evolution from mdfy to memory.wiki shows how branding and positioning can crystallize around a core technical insight (URL as API) to create clearer market positioning

## Key takeaways

- Memory.wiki's core innovation is treating URLs as APIs that any AI can fetch, solving knowledge delivery rather than memory storage
- The three-tier architecture of capture/digestion/utilization with user authoring and AI organization creates a sustainable differentiation from extraction-based competitors
- Strategic focus on cross-AI compatibility leverages a structural advantage that big tech companies cannot replicate due to competitive constraints

## Open questions / gaps

- Competitive analysis depth - limited discussion of how existing tools like Obsidian or Roam Research compare beyond basic positioning
- Technical implementation details - while the vision is clear, specific technical challenges like AI model costs, scaling, and infrastructure are underexplored
- User research validation - the documents are founder-driven without evidence of customer interviews or usage data to validate core assumptions

## Notable connections

- **doc:fz45GXXj** ↔ **doc:UwZWUmdo** — The manifesto's 'delivery vs memory' thesis directly informs the product specification's dual-door architecture and URL-as-API principle
- **doc:xv6A2DKS** ↔ **doc:nvF3Li2x** — The revised plan systematically reduces v7's scope from 8 features to 3 while preserving the core strategic thesis and architecture
- **doc:8QB0olhs** ↔ **doc:Tk71NDw0** — The accessible explanation in the recap directly maps to the formal three-tier architecture specified in the technical document
- **doc:pBEnMXSs** ↔ **doc:xv6A2DKS** — The evolution overview serves as meta-commentary on the finalized business plan, highlighting the strategic transformation process

## Concepts (this bundle)

- **URL as API**
- **Three-Tier Architecture**
- **Knowledge Graph**
- **Cross-AI Compatibility**
- **Delivery vs Memory Problem**
- **User Authoring**
- **LLM Self-Maintenance**

## Concept relations

- **Cross-AI Compatibility** ↔ **URL as API** — enables through
- **Knowledge Graph** ↔ **LLM Self-Maintenance** — built by

## Documents

### 1. [memory.wiki 사업계획 v7 (FINAL)](https://memory.wiki/xv6A2DKS)
> Duplicate business plan v7 with full strategic thesis on knowledge graphs as AI APIs
memory.wiki is a platform that delivers knowledge graphs as URLs and APIs that any AI can use, with the core thesis that "the graph is the product" while maintaining a three-tier architecture of capture, digestion, and utilization of user knowledge.
*sections:* 한 줄로: > memory.wiki — your knowledge graph as a URL for any AI. | The one idea (from product itself): > A memory.wiki URL is an API for any AI. | 핵심 차별화: memory.wiki는 모든 surface가 같은 underlying graph를 읽음:; Concept index (lifelong ontology); Concept relations (typed edges); Bundle graphdata (themes, insights, edges); Embeddings (semantic recall); Mem0/Letta: retrieve memory from extraction; …; LLM Wiki: retrieve knowledge from local compilation; …; memory.wiki: deliver knowledge graph in URL (any AI inherits); … | …

### 2. [Memory.wiki Business Plan and Architecture](https://memory.wiki/pBEnMXSs)
> Architecture synthesis explaining how the knowledge graph serves as a URL-native API for AI systems
Memory.wiki is a platform where users can share knowledge through URLs that function as APIs for any AI system, operating on a three-tier architecture of capture, digestion, and utilization to solve the problem of knowledge delivery across ChatGPT, Claude, Gemini, and similar AI tools. The project rebranded from "mdfy" to "memory.wiki" while maintaining its core principle of making URLs serve as universal APIs for AI knowledge consumption.
*sections:* …

### 3. [내가 memory.wiki를 만드는 이유](https://memory.wiki/fz45GXXj)
> Founder's manifesto on the delivery problem that knowledge graphs solve — AI memory through structured URLs
The real problem with AI memory systems is not that AI cannot remember users, but that user knowledge remains locked in closed platforms like ChatGPT, Notion, and Google Docs where no AI can access it. memory.wiki solves this by providing open URLs that any AI can fetch to access a user's knowledge graph organized as markdown documents, bundles, and hubs.
*sections:* 문제는 memory가 아닙니다. Delivery입니다.: 매일 수백만 명이 ChatGPT, Claude, Gemini, Cursor에 자신의 사고를 쏟아붓습니다. 어려운 질문을 합니다. 정말로 유용한 답변을 받습니다 — 전략, 코드, 프레임워크, 전문가들이 수십 년에 걸쳐 개발한 통찰들. | 단 하나의 아이디어: memory.wiki URL은 어떤 AI든 사용할 수 있는 API입니다. | 세 단계: 수집, 소화, 활용: memory.wiki에는 세 layer가 있습니다. 단순한 workflow로 매핑됩니다. | 그래프가 product입니다: 대부분의 지식 도구는 docs를 저장합니다. memory.wiki는 docs 와 그들 사이의 graph를 저장합니다. | Mem0, Notion, LLM Wiki는 어때요?: memory.wiki가 무엇이고 무엇이 아닌지 정직하게 말하겠습니다. | 왜 지금인가: 이게 중요한 좁은 창이 있습니다. | …


_Digest view — follow any link above to fetch that doc's full markdown. Add `?full=1` to this URL for the concatenated payload._