Liner

Enables users to highlight text on any webpage, which triggers AI-powered semantic analysis to extract key concepts, entities, and relationships from the selected content. The extension integrates with the DOM to capture highlighted regions, sends them to a backend LLM service for contextual understanding, and stores highlights with metadata (source URL, timestamp, semantic tags) in a local or cloud-synced database for later retrieval and cross-referencing.

Provides a search interface within the extension that queries web content and returns answers synthesized from multiple sources, with each claim linked back to its original URL and highlighted passage. The system uses retrieval-augmented generation (RAG) to fetch relevant web pages, extract cited passages, and present them alongside the AI-generated answer, creating a transparent chain from question to source.

Analyzes YouTube video transcripts (auto-generated or manually provided) using NLP to extract key topics, timestamps, and semantic segments, then generates concise summaries organized by theme rather than chronological order. The extension integrates with YouTube's video player to inject a summary panel that links summary sections back to specific video timestamps, enabling users to jump directly to relevant parts.

Aggregates highlighted content, saved sources, and search history into a personalized feed that uses semantic similarity and user interest modeling to surface relevant information. The system tracks which topics the user engages with (based on highlights, searches, and dwell time), builds a user interest vector, and ranks feed items by relevance to those interests using cosine similarity or learned ranking models.

Syncs highlights, searches, and saved content across multiple devices and browsers using a cloud backend with conflict resolution and version control. The system stores highlights with metadata (URL, timestamp, user ID, semantic tags) in a cloud database, implements differential sync to minimize bandwidth, and handles edge cases like duplicate highlights, deleted sources, and offline mode by queuing changes locally until connectivity is restored.

Analyzes the credibility and potential bias of web sources by examining domain reputation, author credentials, publication date, and content patterns using a combination of heuristics and ML models. When a user highlights content or searches, the extension displays credibility indicators (e.g., 'trusted source', 'potential bias detected', 'outdated information') alongside the content, helping users evaluate source quality without manual fact-checking.

Exports highlights in multiple formats (Markdown, JSON, CSV, HTML) and integrates with external tools like Notion, Obsidian, Roam Research, and Evernote via APIs or file-based exports. The extension may support two-way sync with some tools, automatically pushing new highlights to external systems and pulling updates back. Export includes full metadata (source URL, timestamp, tags, color) to preserve context in external tools.

Enables users to share individual highlights or entire highlight collections with teammates, creating shared knowledge bases that multiple users can view, search, and build upon. Shared highlights may be read-only or allow collaborative annotation. The system tracks ownership and permissions (view, edit, comment) and may support team workspaces where highlights are organized by project or topic. Shared highlights are indexed and searchable across the team.