Sign In

What’s new

Updates to NearWord, and the work behind them.

Get the latest version
v0.18 Latest

A clearer lookup and a redesigned website

  • Mac lookup. Select and copy dictionary text, keep your reading position while AI answers arrive, and recover from errors without leaving the lookup panel.
  • Website. Redesigned pages with clearer navigation, mobile layouts, setup instructions, and answers to common questions.
  • Account. Refined sign-in and account settings, clearer billing status, and better recovery after checkout or a payment issue.
v0.17

Downloads and a clearer website

  • Early access. x64 and ARM64 installers for joining the Windows and Linux waitlist. These builds do not include word lookup yet.
  • Mac updates. Fixed an issue that prevented existing installations from finding new releases.
  • Website. Simpler downloads, a clearer Free and Pro comparison, and improved mobile layouts.
v0.16

A native Mac lookup panel

  • AppKit. Native materials, selectable text and a panel that leaves your source app in front.
  • Lookup. AI Context and definitions share one scrollable panel, beside the highlighted word.
  • History. Reopen the explanation and dictionary results from a saved lookup.
v0.15

Smoother sign-in and updates

  • Sign-in. A clearer handoff from your browser back to NearWord.
  • Startup. Choose to open NearWord when you log in to your Mac.
  • Updates. Choose when to install and follow the download progress.
v0.14

The answer comes first

  • Answers. The best match appears above the extra dictionary details.
  • Loading. AI Context and dictionaries load at the same time.
  • Recovery. Clearer error messages, with a retry when AI Context fails.
v0.13

The native app, ready to install

  • Mac downloads. Signed and notarized builds for Apple Silicon and Intel.
  • Updates. Check for a new release and install it from inside the app.
v0.12

A native foundation with GPUI

Tauri · webview GPUI · native Rust
  • Less idle work. Event-driven updates replaced recurring UI timers.
  • Direct capture. Capture and Apple Vision run inside the Mac app, without temporary image files or a helper process.
  • Shared foundation. A common Rust app shell for macOS, Windows and Linux. Platform-specific lookup work continued separately.
v0.11

Your dictionaries, easier to find

  • Library. A dedicated dictionary page, with clearer recent words on the home screen.
  • Feedback. Send a suggestion or report a problem directly from the app.
v0.10

Remember where a word came from

  • Source context. Saved lookups gained app and document details, where available.
  • History. Clearer word cards and language-pair labels made past lookups easier to scan.
v0.9

Your shortcut, your setup

  • Shortcuts. Record a key combination in Settings instead of using the default.
  • Startup. Dictionary and recognition assets prepare in parallel, with progress shown as they load.
v0.8

A growing dictionary library

  • Dictionary packs. Introduced curated dictionary data and a versioned download catalog.
  • Under the hood. A dedicated builder turned source dictionaries into consistent offline packs.
v0.7

More context for better explanations

  • AI answers. Refined explanations around the surrounding text, with clearer translations.
  • Retry. Try an AI lookup again when a request fails, without starting over.
Engineering

From a prototype to an installable app

  • Distribution. Added macOS signing and notarization, with direct downloads from the website.
  • Recognition. Reduced copying in the OCR image pipeline and added screen-capture regression tests.
Experiment

Trying AI on the device

  • The experiment. Tested small language models running locally for offline explanations.
  • What we learned. Answers weren’t consistent enough for everyday reading, so we removed the experimental mode. On-device AI remains on our roadmap.
Engineering

Our own OCR pipeline

  • Recognition. Built a Rust pipeline around PaddleOCR models and ONNX Runtime, with our own image processing and multilingual tests.
  • Today. The Mac app uses Apple Vision. The standalone engine remains a foundation for Windows and Linux work.
Engineering

A history that stays with you

  • Local storage. Moved lookup history into SQLite, with compressed thumbnails for revisiting earlier words.
  • Maintenance. Added update checks and an in-app installation flow to the Tauri app.
Engineering

A website, an API and contextual answers

  • Connected product. Built the SvelteKit website and connected the app to an API for contextual explanations.
  • AI responses. Streamed answers into the lookup panel as they arrived.
  • Platform research. Built early Windows versions and explored Linux capture. These were not ready for everyday lookup.
Engineering

A shared app with Tauri

  • Architecture. Paired a Rust backend with a Svelte interface to explore a cross-platform app.
  • The core interaction. Brought screen capture and the lookup shortcut into that shared foundation.
Prototype

One word, one shortcut

  • SwiftUI. Started with a small Mac app. Point at a word, press a shortcut and see its meaning without leaving what you’re reading.

NearWord

Since 2025