lbstrydom406 downloadsYour AI thinking partner in Obsidian: summarize anything, research the web, build slide decks, transcribe meetings, and auto-tag your vault. Cloud or local LLMs — your own API key.
Your notes deserve more than storage. AI Organiser turns Obsidian into a thinking partner — summarize anything, research the web, build presentations, record meetings, sync Kindle highlights, and let AI tag and connect your knowledge. One plugin, 14 providers, works locally or in the cloud.

Drop in a URL and get a structured summary with tags — or paste five URLs and get a single synthesized note. Record a meeting and walk away with structured minutes, action items in GTD format, and a Word document ready to send. Ask "what do I know about X?" and get answers grounded in your own vault, with citations. Say "make me a deck about Q3 results" and watch slides appear in a live preview, then export to PowerPoint.
Everything works from a single Command Picker — press one hotkey, browse by category, search by keyword. No menus to memorize.
Chat about your active note, search your vault with semantic RAG, research the web with source scoring and citations, build slide decks in conversation, or just talk. Conversations persist across sessions, support file attachments, and remember context through projects and global memory.
URLs, PDFs, YouTube videos, audio recordings, Office documents — individually or in bulk. Choose from 5 built-in personas (Student, Executive, Technical, Researcher, Casual) or write your own. Every write shows an inline diff preview so you approve changes before they land.
Ask a question and the plugin decomposes it into sub-queries, searches multiple providers, scores source quality, extracts findings, and synthesizes an answer with numbered citations. Academic mode adds DOI extraction and author-year formatting. Budget guardrails stop you from accidentally burning through your API credits.
Describe what you want in plain language and watch slides appear in a live, sandboxed preview with a side-rail workspace and thumbnail filmstrip. Chat to refine the whole deck or polish individual slides. Export to editable PowerPoint — native shapes, tables, and charts (2×2 matrices, Harvey balls, waterfalls, line charts, pyramids), not flat images — or to self-contained HTML.
Turn on consultant-quality mode to get a McKinsey/BCG-style workflow: the plugin first drafts a grounded, MECE storyline (action titles, evidence-bound visuals) that you review and sign off on before any slide is designed. Optional brand fidelity reads a brand pack from your vault (colours, fonts, logos, icons) so both PPTX and Word exports come out on-brand and template-ready.
Record audio in-plugin or paste a transcript. Get structured minutes with agenda items, decisions, and action items — optionally classified by GTD context (@office, @call, @errand). Terminology dictionaries keep names and acronyms consistent. Export to Word with one click.
Connect your Gmail via a simple Apps Script. The plugin fetches unread newsletters, triages each one with AI, writes individual notes with key links, and synthesizes a thematic daily brief that groups stories across sources. Optional audio podcast generation reads you the brief.
Turn any note into a spoken-audio MP3 — "read this note to me" — with an optional LLM pre-pass that summarises diagrams and large tables and expands acronyms for natural listening. A built-in player offers speed and skip controls. Cost is shown and confirmed before any API call.
Auto-tag notes using your own taxonomy with 3-tier hierarchical tags. Match existing tags, generate new ones, or use a hybrid. Batch-tag entire folders. Visualize your tag network as an interactive D3.js graph.
7 embedding providers including a zero-setup local option (no API key needed). Find related notes in a persistent sidebar. Export search results as new notes. Every AI response can be grounded in your vault context.
Works with 14 cloud providers and 5+ local options — pick what fits your budget and privacy needs.
| Provider | Example Models |
|---|---|
| Anthropic | Claude Opus 4.6, Sonnet 4.6, Haiku 4.5 |
| OpenAI | GPT-5.4, GPT-4o, o3, o4-mini |
| Google Gemini | Gemini 2.5 Pro, 2.5 Flash |
| DeepSeek | R1, V3 |
| Groq | Llama, Mixtral (fast inference) |
| Mistral | Large, Codestral |
| Cohere | Command R+ |
| Grok | Grok-2 |
| OpenRouter | 200+ models |
| Requesty | Multi-provider routing |
| Amazon Bedrock | Claude, Llama, Titan |
| Google Vertex AI | Gemini models |
| Alibaba (Aliyun) | Qwen models |
| SiliconFlow | Open-source models |
Local: Ollama, LM Studio, LocalAI, Jan, KoboldCpp
Enterprise: Azure AI Foundry — Claude + GPT + embeddings + Whisper under one resource, with flexible per-capability routing (use an Azure deployment, bring your own provider, or off) so any corporate Foundry works without silent fallbacks.
Embeddings: OpenAI, Gemini, Ollama, Cohere, Voyage AI, OpenRouter, Local ONNX (zero-setup, no API key)
Works on iOS and Android — cloud providers, touch-optimized modals, audio recording with automatic codec negotiation, RAM-aware indexing.
English interface. Output language is configurable independently — generated content (summaries, tags, minutes, translations) can target 20+ languages.
main.js, manifest.json, and styles.css from the latest release<vault>/.obsidian/plugins/ai-organiser/See docs/architecture-map.md — a generated
index of every symbol in this repo, grouped by domain, with Mermaid
diagrams. Regenerated on /ship or via npm run arch:render.
git clone https://github.com/Lbstrydom/ai-organiser.git
npm install
npm run dev # watch mode
npm test # 5700+ unit tests
npm run build # production
See AGENTS.md for architecture documentation.
GPL-3.0 — see LICENSE.