Skim Reader

The tl;dr for the modern doomscrolling world. An RSS reader built around one interaction: a haptic slider that compresses any article to match the time you have. A 30-second overview or the full piece, with a single gesture.

iOS Released
Download Skim Reader on the App Store

Version 1.1. Free with optional AI model download.

Article view at the Glance level showing a two-sentence summary
Glance view. Two sentences, 10 seconds to read
Article view at the Full level showing the complete article text
Full article, when you have the time
Article list for a feed with headlines and excerpts
Article list
Feed management screen listing popular feeds to add
Feed management, with popular feeds to get you started

What it does

  • The compression slider

    A haptic slider with 5 tiers, from a 10-second glance to the full article. Each tier snaps with haptic feedback.

  • On-device AI summaries

    A locally-running LLM generates summaries on your device. No cloud, no servers, no data leaving your iPhone or iPad.

  • Adaptive learning

    The app learns your preferences per feed and time of day, then opens articles at your preferred compression level.

  • Typography

    New York display font, Charter body text and user-selectable typefaces.

  • Bionic reading

    An optional mode that bolds the first portion of each word, creating fixation points for faster scanning.

  • Organisation

    Swipe to bookmark or mark read. An All Articles view aggregates your feeds. Hide read articles.

  • Designed for iPad

    Split-view navigation with a sidebar. Feeds and articles side by side, with keyboard shortcuts.

  • Save any URL

    Paste an article URL to summarise it without subscribing to a feed. Useful for one-off reads.

The compression slider

Five discrete positions with haptic feedback at each snap. From a glance to the full article.

T1 Glance
1 to 2 sentence headline summary. About 25 words, 10 seconds.
T2 Brief
Key points at a glance. About 100 words, 30 seconds.
T3 Summary
A fuller overview. About 400 words, 2 minutes.
T4 Digest
Detailed, with nuances preserved. About 800 words, 5 minutes.
T5 Full
The entire original article. Length varies.

Privacy

Your reading habits are nobody's business. The LLM runs entirely on your device, so your articles and summaries never leave your iPhone or iPad. No servers, no cloud.

All feeds, articles and cached summaries are stored on your device with SwiftData.

We don't track what you read, when you read, or what you subscribe to.

The AI model

Skim Reader uses a quantised Qwen2.5-1.5B-Instruct model running locally via llama.cpp. The model is downloaded once during onboarding (about 1.12 GB) and runs on your Apple Silicon chip with Metal GPU acceleration.

  • 5 to 10 seconds per summary
  • Summaries are cached, so they open instantly the second time
  • Works offline after the initial model download

Requirements

  • iOS 26.0 or iPadOS 26.0 or later
  • iPhone or iPad with an A17 chip or later, for the best LLM performance
  • About 1.2 GB of storage for the AI model
  • Internet for feed refresh and the initial model download

How it works

  1. Add feeds

    Enter any URL. The app discovers RSS feeds from websites, or you can add direct feed URLs.

  2. Open and slide

    Open an article and drag the compression slider. Each tier snaps with haptic feedback.

  3. Let it learn

    The app learns your preferences over time and opens articles at your preferred depth.