Readwise Reader

Consolidates articles, newsletters, PDFs, tweets, YouTube transcripts, and EPUB ebooks into a single centralized database through browser extension highlighting, direct uploads, and upstream integrations (RSS, email forwarding, social media). Content is normalized into a common schema with metadata (source, timestamp, tags, notes) and indexed server-side for subsequent AI processing and retrieval.

Enables users to ask natural language questions about saved documents and highlights using GPT-4 as the underlying model. The system retrieves relevant document context, constructs a prompt with the user's question and document text, and returns GPT-4's response. Implementation details (prompt engineering, context window management, token limits) are not publicly documented.

Automatically extracts transcripts from YouTube videos when users save video URLs to Readwise, making transcripts available for highlighting, searching, and AI processing. Extraction uses YouTube's native transcript API (if available) or third-party transcript services. Extracted transcripts are indexed and associated with video metadata (title, channel, duration, upload date).

Enables users to save Twitter threads and individual tweets to Readwise, extracting thread content (tweets, replies, author metadata) and making them available for highlighting and searching. Threads are preserved as complete units with conversation context, protecting against tweet deletion or account suspension.

Supports RSS feed subscriptions and email newsletter forwarding, automatically ingesting new articles and emails into the Readwise library. Feed items are normalized with metadata (publication date, author, feed source) and made available for highlighting, searching, and AI processing. Newsletter forwarding uses a unique email address per user.

Automatically generates summaries of saved articles, newsletters, and documents using an unspecified AI model (not documented as GPT-4). Summaries are computed server-side and presented alongside the original content. Implementation approach (extractive vs. abstractive, model architecture, summary length configuration) is not publicly disclosed.

Implements a proprietary spaced repetition algorithm (branded as 'Daily Review') that selects highlights from the user's collection and resurfaces them at optimal intervals based on cognitive science principles. The system tracks highlight review history, calculates optimal review timing, and delivers a curated daily digest via email or in-app interface. Algorithm details (interval calculation, decay function, weighting factors) are not publicly documented.

Enables keyword and semantic search across all saved highlights and documents in the user's Readwise library. Search indexes full-text content from articles, PDFs, newsletters, and other sources, returning results with source attribution and highlight context. Implementation approach (inverted index, vector embeddings, hybrid search) is not documented.

Provides a browser extension that enables users to highlight text on web pages and attach notes directly within the reading context. Highlighted content is captured with metadata (URL, timestamp, selection context) and synced to the centralized Readwise database. Extension integrates with the browser's native text selection and context menu APIs.

Accepts PDF and EPUB file uploads, extracts full text content server-side, and makes documents available for highlighting, searching, and AI processing (summarization, Q&A). Extraction handles multi-page documents, embedded images, and complex layouts. Extracted text is indexed and associated with document metadata (filename, upload date, file size).

Enables bulk export of highlights and notes to external platforms (Notion, Obsidian, Evernote) via API integrations or formatted exports. Export includes highlight text, source attribution, tags, notes, and metadata. Integration maintains bidirectional sync or one-time export depending on platform.

Provides a hierarchical tagging system allowing users to organize highlights and documents with custom tags, enabling filtering, search, and organization. Tags are user-defined, can be applied to individual highlights or entire documents, and are searchable. System supports tag hierarchies and bulk tag operations.

Allows users to attach freeform text notes to individual highlights, providing context, interpretation, or follow-up thoughts. Notes are stored alongside highlights, searchable, and exported with highlight data. Notes support basic formatting (unknown if markdown or plain text only).