Aether vs Wikilayer

Aether turns every link you save into structured knowledge. Drop any URL — YouTube videos, Instagram posts, articles, research papers, blog posts — and AI automatically extracts, transcribes, analyzes, and summarizes the content into your personal searchable vault. HOW IT WORKS: 1. Share a link from any app (YouTube, Instagram, Safari, Chrome) 2. AI extracts content — transcribes videos, OCRs images, scrapes articles 3. Get rich AI-generated summaries with key insights and entities 4. Search your entire vault by meaning, not just keywords KEY FEATURES: • Instant Capture — Share from any app via iOS Share Extension • YouTube Transcription — Gemini AI transcribes video audio • Instagram Vision — AI analyzes carousel images with OCR • AI Summaries — Claude generates structured insights • Semantic Search — Find anything by meaning using vector embeddings • Auto-Labels — Content auto-organized by source (YouTube, Instagram, GitHub, etc.) • Bilingual — Full English and Turkish language support • Dark Mode — Beautiful obsidian dark interface SUPPORTED SOURCES: YouTube, Instagram, Twitter/X, Reddit, GitHub, Medium, Wikipedia, arXiv, Stack Overflow, Hacker News, and any web page. Your knowledge, always searchable, always yours.

Wikilayer is a wiki that an AI assistant can read and edit on your behalf. You and your agent work on the same tree. The agent writes through MCP tools: it adds a node, rewrites one, moves a subtree, and reads back by outline instead of loading everything into its context. You read the result here, as ordinary pages. Every block has an address of its own. Rename it, rewrite it, move it under another parent, and links to it still land. Pages that lead to a block are listed under it, and nobody maintains that list by hand. History is kept per node, so a change to one paragraph is a change to one node, not to a whole document. WHAT PEOPLE KEEP HERE Personal notes: a second brain the agent reads selectively instead of holding a whole vault in context. Adding is a message in chat, not an app switch. Project memory: long-running context the agent writes and you audit. Hierarchy and full-text search across titles and bodies, without a separate vector store. Team workspace: decisions, context and meeting digests, written by the team and refactored by the agent. Public reference: guides, handbooks, fan wikis. Readers arrive from search and never touch the editing surface. VISIBILITY A wiki can be open to everyone, private to a team, or run on your own machine. Public wikis are crawlable the day you write them: server-rendered pages, Open Graph, sitemap, structured data, and addresses that survive a rename. WHAT IT IS NOT Not a block editor: bodies are plain markdown edited through MCP, not click-to-edit blocks, and the tree is hierarchical nodes rather than a relational document store. Not a real-time co-editor: each edit lands atomically, and conflicts resolve at write time rather than by merging keystrokes. Not an approval workflow: edits land on the live tree, and the audit trail is a snapshot of each node as it stood before the change. Bodies are markdown and the tools are MCP. Both are public formats, so a wiki reads out as plain documents whenever you want it somewhere else.