Notes & knowledgeCAN I VIBECODE
Fabric?
BUILD FIRST
- One specific customer type and one primary use case
- A vertical slice that completes the main job without manual workarounds
- Visible limits, review steps, and a fallback when an external dependency fails
LEAVE FOR LATER
- Broad third-party integrations
- High-scale collaboration, syncing, and permissions
- Automation that can cause harm without human review
RESEARCH NOTES
What this tool actually does.
Fabric (fabric.so, by Fabric Internet Co.) is an AI-powered, all-in-one notes and knowledge workspace marketed as "a personal AI that actually knows you." It unifies notes/docs, files, ideas, meeting notes, and voice memos in one place for ~300k+ users. Core features per the official site: a Markdown-native notes editor supporting templates (meeting notes, book notes, reviews), rich linking, and embedded content; "deep search" that finds content by meaning across notes, PDFs, articles, and voice memos; an AI assistant that answers questions about your library with cited sources; agents; tasks; canvas; publishing; and email-to-Fabric capture. It claims 50+ integrations (Gmail, Dropbox, Google Drive, Notion), multiplayer real-time editing with comments and @mentions, sync across iOS, Mac, web, and Android, SSL/AES-256 encryption, AI transcription of voice notes, and importers for Raindrop,
RESEARCHED BUILD PROMPT
Give this to your coding agent.
Build a focused, smaller replacement for Fabric (fabric.so), a personal AI notes & knowledge workspace ("second brain"), under a different name with original UI and branding. Core workflow to replicate: capture notes, links, voice memos, and files from one quick-capture surface; index everything; search by meaning, not just title; chat with an assistant that answers across the whole library with citations; organize via Markdown notes, rich linking, tags, and templates. User-facing requirements: Markdown editor with note-to-note links and embedded images/files; quick-capture (a launcher-style web clipper and a voice note recorder with AI transcription); semantic search over notes plus uploaded PDFs/txt files; a chat assistant whose answers cite source snippets; per-note templates; sync across devices with auth. Technical requirements: TypeScript monorepo; React (Next.js) front end; Postgres for metadata; pgvector embeddings for semantic search; an OpenAI-compatible embeddings/chat API (with a configurable local model option); object storage for files; a background worker queue for indexing and transcription; WebSocket collaborative editing via CRDT (Yjs) so two users can edit a note live. Aim for a working vertical slice: sign up, capture a note and a voice memo, search by meaning, ask the assistant a question grounded in your notes. Do NOT copy Fabric's brand, logo, marketing copy, screenshots, or UI layout. Deliberately out of scope: native mobile apps, the 50+ third-party integrations marketplace, meeting-bot transcription, publish/portals, tasks/agents, templates marketplace, billing, and offline mode. Deliver a README covering setup (env vars, Docker or local Postgres, migrations, worker start), architecture overview, and important limitations (e.g., semantic search quality depends on the embedding model, no offline or PWA sync, single-instance deployment).