LLM Wiki Systems
A curated list of AI-powered wiki platforms, knowledge management tools, and LLM-driven knowledge base systems. Last updated: 2025.
Wiki Platforms with AI
| Project | Company | Stars | Open Source | Description |
|---|---|---|---|---|
| Outline | Outline HQ | 29k+ | BSL 1.1 | Team wiki with OpenAI/Claude/Ollama support, semantic search, AI writing assistant, and RAG Q&A. |
| AppFlowy | AppFlowy Inc. | — | AGPL | Open-source Notion alternative. Local-first, supports LLM plugins. |
| Docmost | Community | — | AGPL | Emerging open-source wiki similar to Confluence/Notion. AI features on the roadmap. |
| Wiki.js | Requarks | — | AGPL | Modular wiki platform. No native AI; extensible via APIs. v3.0 roadmap includes AI features. |
| BookStack | Dan Brown | — | MIT | Simple self-hosted documentation wiki. No native AI; API available for LLM integration. |
| Confluence | Atlassian | — | Commercial | Enterprise wiki standard. Atlassian Intelligence + Rovo AI: AI summaries, natural language search. |
| Notion AI | Notion Labs | — | Commercial | All-in-one workspace. AI layer: writing, summaries, Q&A, database auto-fill, AI Agents (2025). |
Enterprise Knowledge Platforms
| Product | Company | Open Source | Description |
|---|---|---|---|
| Onyx (Danswer) | Onyx (YC W23) | MIT (community) | Connects 50+ sources (Confluence/Notion/Slack). Self-hosted RAG enterprise Q&A, any LLM. ⭐13k+ |
| Glean | Glean Technologies | No | Enterprise AI search across 100+ integrations. Builds org knowledge graph with permission-awareness. |
| Guru | Guru Technologies | No | AI knowledge management. Guru AI answers questions in real time from internal docs. Slack/Teams integration. |
| Tettra | Tettra | No | Slack-native internal wiki. Kai AI answers directly in Slack. Good at knowledge gap detection. |
| Document360 | Kovai.co | No | Internal/external docs platform. Eddy AI: search Q&A and embedded customer support chatbot. |
| Slite | Slite | No | AI-driven document management. Auto-tagging, search, summaries. Good for small/mid teams. |
Personal Knowledge Management (PKM) with LLM
| Project | Company | Stars | Open Source | Description |
|---|---|---|---|---|
| Logseq | Logseq | 30k+ | AGPL | Outline + graph PKM. DB version supports stronger AI integration; community plugins for LLM. |
| SiYuan Note | Community | 20k+ | AGPL | Block-level editor, local-first. Community plugins support AI summaries and RAG. |
| Foam | Community | — | MIT | Roam Research-style PKM inside VS Code. LLM integration via Continue.dev. |
| Trilium Notes | Community | — | AGPL | Hierarchical self-hosted personal knowledge base. Community LLM integration scripts available. |
| Obsidian | Obsidian | — | Partial (plugins MIT) | Local-first Markdown knowledge graph. Smart Connections / Copilot plugins support Ollama/OpenAI RAG. |
| Mem.ai | Mem Labs | — | No | AI-native notes. Mem X auto-organizes and links notes. Fully LLM-driven. |
| Capacities | Capacities | — | No | Object-type PKM. LLM writing assistant and intelligent search (2025). |
| Google NotebookLM | — | No (free) | LLM-native notebook. Upload docs → auto summaries, podcast generation, Q&A. Deep Gemini integration. |
Agent Memory Frameworks
Systems focused on giving AI agents persistent, cross-session memory — distinct from document Q&A.
| Project | Company | Stars | Open Source | Description |
|---|---|---|---|---|
| Cognee | Community | — | Apache 2.0 | Open-source AI memory platform. Auto-builds knowledge graphs; remember / recall / forget / improve API. Vector search + graph reasoning for cross-session Agent memory. |
| Mem0 | Mem0 AI | — | Apache 2.0 | Persistent memory layer for AI agents and assistants. Stores and retrieves user preferences and context across sessions. |
| Letta | Letta AI | — | Apache 2.0 | Stateful agents with persistent memory backed by Neo4j. Formerly MemGPT. |
| MemOS | MemTensor | — | Apache 2.0 | Memory operating system for agents. Active memory management, hybrid retrieval, skill evolution. MemReader model for memory extraction. |
| Engram | Community | — | MIT | Lightweight, local-first persistent memory system for AI coding agents. Alternative to Mem0+Qdrant or Letta+Neo4j. |
Research & Specialized Projects
| Project | Company | Stars | Description |
|---|---|---|---|
| WikiChat | Stanford OVAL | — | Uses Wikipedia to reduce LLM hallucination. Retrieves and verifies before generating answers. |
| LLM Wiki | nashsu (community) | — | Desktop app (Tauri) inspired by Karpathy's LLM Wiki idea. Incrementally builds a persistent, interlinked wiki from your documents. Obsidian-compatible output. "Human curation, LLM maintenance." |
| GBrain | Garry Tan (YC CEO) | — | Knowledge operating system. Handles multi-source, multimedia input (audio transcription, etc.) and continuously updates a structured knowledge graph. Built for Agent long-term memory. |
| Sage Wiki | Community | — | Go single-binary LLM Wiki implementation. Tiered compilation (supports 100k+ docs), hybrid search (FTS + vector + ontology graph), 17 MCP tools. Cuts token usage by up to 95% vs naive RAG. |
| Walnut | Community | — | Local-first AI knowledge management Agent. BYOK mode (OpenAI/Claude/any provider). Auto-organizes docs and proactively recommends related notes during writing. |
| PieKBS | — | — | Local-first knowledge search engine for agents. Distills raw docs into structured Markdown wiki; search and read via MCP. |
| Mintlify | Mintlify | — | Auto-generates developer docs/wiki from code. AI-driven documentation platform. |