Search Everything.
Locally.
One command palette to search your entire knowledge stack in parallel. Notes, documents, wikis, the web, and your local LLM — all from a single hotkey. Zero cloud dependency.
The Omnilocal Search
One query. Seven sources. Zero latency for cached results. Type once and every knowledge source in your local stack lights up simultaneously.
Vector cache — instant semantic matches from previous searches
InstantScanned documents, receipts, PDFs with full OCR text
LiveWiki pages, shelves, chapters — your documentation hub
LiveMarkdown notes, daily journals, knowledge graph links
LiveSearXNG or Brave — private web results without tracking
LiveLocal LLM chat — ask questions, get AI answers on-device
LiveRAG knowledge base — answers grounded in your documents
LiveQdrant cached results appear instantly. Live sources stream in as they complete.
Everything From One Palette
Not just search. Prism is a full command interface for your local infrastructure.
Omnilocal Search
Query 7 sources in parallel with a single keystroke. Qdrant vector cache returns semantic matches instantly while live sources stream results.
Command Palette
Type > to trigger n8n workflows with parameter dialogs. Run automations without leaving the keyboard.
Local LLM Chat
Streaming conversations with your local LLM. Conversation branching lets you explore multiple lines of thought from any message.
Screenshot OCR
Capture any region of your screen, extract text with OCR, and feed it directly to your local AI for analysis or summarization.
Smart Prompts
Type / to browse categorized prompt templates. Coding, writing, analysis — all a keystroke away.
Global Hotkey
Summon Prism from anywhere with Ctrl+Alt+Space (Windows/Linux) or Cmd+Option+Space (macOS). Instant access, always.
Search History & Pinning
Every search is logged in local SQLite. Pin important results for quick access. Your search history is yours alone — never uploaded.
Webhook Automation
Fire webhooks on search, chat, or OCR events. Connect Prism to n8n, Zapier, or any HTTP endpoint to trigger downstream workflows.
Your Searches. Your Data. Your Hardware.
Prism was built with a single conviction: your knowledge should never leave your network. Every search, every chat, every OCR capture stays on your local disk. No telemetry. No analytics. No cloud calls.
- No telemetry or usage analytics — ever
- No cloud API calls — all AI runs on your hardware
- No tracking pixels, no fingerprinting, no cookies
- Works fully offline — airplane mode ready
- All data stored in local SQLite — portable and inspectable
- Open source Tauri 2 + Rust backend — audit the code yourself
Connects to Your Self-Hosted Stack
Prism speaks to the services you already run. Every integration points at your own infrastructure — no third-party SaaS required.
Qdrant
High-performance vector database for semantic search. Caches embeddings from all sources for instant recall of previous results.
Self-HostedPaperless-ngx
Document management with OCR. Search scanned receipts, invoices, contracts, and any PDF in your archive.
Self-HostedBookStack
Self-hosted wiki platform. Search across shelves, books, chapters, and pages of your team documentation.
Self-HostedObsidian
Markdown-based knowledge base. Search notes, daily journals, and follow backlinks through your personal knowledge graph.
Local VaultSearXNG / Brave
Privacy-respecting web search. Get real web results without Google tracking. Self-host SearXNG or use Brave Search API.
Self-Hosted / APILM Studio
Run open-weight LLMs locally. Streaming chat, completions, and embeddings on your own GPU. Llama, Mistral, Phi, and more.
LocalAnythingLLM
RAG-powered knowledge base. Upload documents and get AI answers grounded in your actual files. Fully self-hosted.
Self-Hostedn8n Workflows
Trigger any n8n workflow from the command palette. Pass parameters, receive results, automate your entire homelab from one hotkey.
Self-HostedAvailable on Every Desktop
Built on Tauri 2 for native performance. Lightweight, fast, and cross-platform.
30-day free trial. No credit card required. No account needed.
Take Back Your Search.
Stop sending your queries to the cloud. Search your entire local knowledge stack from one command palette. Privacy by design.
GitHub · Duncan Smith · Blog