Features

Capture From Everywhere (Without Changing Your Workflow)

ONA.UNO brings your information into a single timeline — without moving your originals.

• Files and folders — Index documents where they live: text/markdown, PDFs, images, and HTML.
• Browser Clips (Chrome/Chromium + Safari) — Save any webpage in one click/shortcut; ONA.UNO converts it to clean Markdown and deduplicates by canonical URL.
• Apple Mail — Index your Mail archive from local storage (no OAuth); choose accounts and date ranges.
• Obsidian Vault — Markdown-first ingestion with frontmatter support (titles, tags, aliases) and deep links back to obsidian://.
• Feedbin (Starred) — Your starred articles, imported with full text and kept in sync.

Everything lands in one place: a chronological timeline you can browse by day, tag, or source.

A Sophisticated AI Pipeline (That Runs Automatically)

Every item that enters ONA.UNO is enriched in the background:

• Clean extraction (readability-based extraction for HTML, text extraction for PDFs, OCR for scanned PDFs and images)
• YouTube transcripts when you clip YouTube videos (so spoken content becomes searchable)
• Chunking + embeddings (so meaning, not just keywords, becomes searchable)
• Micro summaries (short, scannable bullet points)
• AI titles (descriptive names generated from content)
• AI tags (topic labels you can click and browse)

The result: you don't "file" things anymore — your knowledge organizes itself.

Summaries At Every Level

Scan fast when you're busy. Go deep when you need detail.

• Micro summaries are generated automatically for each item (short, scannable, consistent).
• Long summaries can be generated on demand for a day, a time range, a tag, or a search result.
• Summaries can include citations so you can trace every claim back to its source.

Search That Actually Finds Things

ONA.UNO is built for libraries with thousands (or tens of thousands) of items.

Choose the right tool for the job:

• Titles — fast when you remember the name (includes AI-generated titles).
• Titles, Tags & Summaries — full-text search across the "important" fields.
• Titles, Tags & Full Content — search everything, including full note content.
• Semantic (AI) — meaning-based search for "what was that thing about..." moments.

Power users get a real query language:

• tag:"project alpha", title:roadmap, is:starred, is:youtube
• day:today, day:week, date>=2025-01-01
• AND, OR, NOT, parentheses

Chat With All Your Knowledge (With Citations)

Sometimes you don't want to browse — you want to ask.

ONA.UNO chat is designed for serious, traceable answers:

• Chat with all items in your library
• Chat with a focused scope: a selection, a day or time of day (morning, afternoon, evening), a tag, or your current search results
• Citations by default — answers include [#] references you can click to jump back to the source
• Open original — jump from a citation directly into Obsidian, Finder, Mail, or the web page
• Chat history — conversations are saved, reloadable, and renameable

This is not a generic chatbot. It's a chat interface grounded in your sources.

Two AI Modes: Online and Mixed

ONA.UNO gives you a clear choice depending on your hardware, costs, and privacy preferences.

Online Mode (Best Quality, Fastest Bulk Imports)

Online mode runs the full pipeline via OpenRouter:

• Best-of-class embeddings and pipeline models (Qwen3 Embedding 8B + Gemini 2.5 Flash Lite)
• Parallel processing for faster initial indexing
• Works on Intel Macs and Apple Silicon

Mixed Mode (Cost-Effective, More On-Device)

Mixed mode runs high-volume pipeline stages locally (on Apple Silicon), while still using the best remote models when it matters:

• Local embeddings + micro summaries + titles + tags (mxbai-embed-large + Qwen 2.5 7B)
• Remote chat + long summaries (Gemini 2.5 Flash Lite or Grok 4.1 Fast)
• Great for lower ongoing costs and offline pipeline processing

In both modes, you always see what's happening: ONA.UNO tracks remote spend (month-to-date + last request) so there are no surprises.

Extensions and Integrations That Disappear Into Your Workflow

ONA.UNO is built to meet you where you already work.

Chrome + Safari Clipper

• One click, shortcut, or context menu: Save to ONA.UNO
• Converts pages to clean Markdown (excellent extraction quality)
• Offline queue: clip now, imports later automatically when ONA.UNO is running
• Smart deduplication by canonical URL
• Export and import your clips as Markdown (useful for transfers and backups)

Obsidian Companion Plugin

Optional, but powerful:

• Open in ONA.UNO from the current note
• Chat in ONA.UNO about the current note
• Status indicator showing whether ONA.UNO can see your vault
• Works locally (no network required)

Raycast Extension

Fast access without switching contexts:

• Search Items — full-text search with live preview; open results in ONA.UNO or your browser via Web Access
• Recent Items and Recent Starred Items — jump straight to what you saved or starred
• Clip Item — send a URL to ONA.UNO for clipping
• Chat with All Items — start a new chat session instantly
• Daily Summary — read today's AI-generated summary without leaving Raycast

Two connection modes:

• Local Mode — connects directly to ONA.UNO on your Mac
• Remote Mode — connects via Web Access to ONA.UNO on another Mac (great with Tailscale)

Web Access (Local-First, Remote via Tailscale)

ONA.UNO includes an integrated web server ("Web Access") so you can use your library from any browser — phone, tablet, or another computer.

• Browse your timeline and open item pages
• Run search (including semantic search)
• Use chat with streaming answers and clickable citations
• Clip a URL from mobile (no extension needed)

Security model:

• Off by default, local network only when enabled
• Password-protected login
• Optional remote access via Tailscale (encrypted, no port forwarding, easy setup)

Native macOS Performance

ONA.UNO is a native macOS app built for large libraries:

• Smooth scrolling and fast UI, even with thousands of items
• Search designed to stay responsive (you hit Enter to search, no sluggish "live search")
• Background processing that doesn't block your work

Your Data, Your Control

• Your originals stay where they are — ONA.UNO indexes, it doesn't take over your files.
• There is no ONA.UNO server that stores your content.
• Cloud AI usage is BYOK ("bring your own key") via OpenRouter, with hard spending limits you control.
• Your content goes directly from your Mac to the AI provider over encrypted connections — we never see it.