Valin

Product Introduction

1. Definition: Valin is a private, on-device AI bookmarking and memory application designed for iOS and iPadOS. Functioning within the personal knowledge management (PKM) category, it acts as a localized search engine and semantic synthesizer for cross-platform content.
2. Core Value Proposition: Valin solves the "digital black hole" of modern bookmarking by ingesting videos, posts, articles, PDFs, and voice notes, then running local AI models to extract information down to spoken audio and video frames. Its primary value proposition centers on absolute data privacy—operating entirely on the local device without requiring a cloud account, external AI subscriptions, or data tracking.

Main Features

1. Multi-Modal Content Processing: Valin parses multi-format saves by reading on-screen text from video frames and images, transcribing spoken audio via Whisper (Small via WhisperKit, ~489 MB), and extracting textual context from captions and PDFs. The system compiles takeaways, summaries, dimensions, and instructions into a single unified library item.
2. Local Semantic Search and Ask: Powered by EmbeddingGemma (300M, ~349 MB), the search engine indexes items by meaning rather than exact keywords, allowing users to query information using natural language. The "Ask" feature uses Gemma 4 (E2B, ~2.59 GB) to synthesize answers across multiple saved documents, providing numbered citations and source cards linked directly to the original context.
3. Automated Collections and Recipe Extraction: Valin analyzes saved items to automatically suggest thematic collections for projects or learning subjects. Additionally, its dedicated recipe extraction tool isolates ingredients, precise quantities, and step-by-step instructions from cooking reels and articles, eliminating the need to manually pause and rewind video content.

Problems Solved

1. Pain Point: Digital hoarding across fragmented applications (Instagram, TikTok, YouTube, Safari, Notes), leading to content loss, inaccessible deep details trapped inside video/audio streams, and privacy breaches inherent in cloud-based AI indexing tools.
2. Target Audience: Researchers, students, content creators, productivity enthusiasts, and privacy-conscious professionals who heavily clip web articles, social media posts, and multimedia tutorials on iOS devices.
3. Use Cases: Retrieving a specific DIY instruction mentioned verbally at a precise timestamp in a TikTok or Instagram Reel; comparing multiple saved interior design or product options through multi-source AI queries; extracting recipe steps from saved cooking videos; and reading saved PDFs offline with automated summaries and table of contents generation.

Unique Advantages

1. Strengths & Limitations (Pros & Cons):
   • Pros: Complete data sovereignty with 100% on-device AI processing; no registration walls or recurring SaaS subscriptions; offline functionality for search, reading, and summaries once models are downloaded; multi-modal parsing that successfully extracts data from audio and video frames.
   • Cons: High hardware requirements (iOS/iPadOS 26 or later, device with at least 8 GB of RAM including iPhone 15 Pro, iPhone Duo equivalents, and compatible iPads); substantial local storage consumption (~3.43 GB for AI models plus cache and media); accelerated battery drain during active on-device processing; exclusion of content locked behind private walls or strict paywalls.
2. Key Alternatives & Differentiation:
   • Readwise Reader / Omnivore: While excellent for cross-platform article and ebook reading, traditional bookmarking tools rely heavily on cloud infrastructure and typically lack advanced on-device video frame analysis and local audio transcription capabilities.
   • Standard Cloud AI Notebooks (e.g., NotebookLM): Offer robust multi-document synthesis but require uploading personal data to external servers, violating strict privacy requirements. Valin differentiates itself by executing local models via Gemma and Whisper, ensuring data never leaves the user's hardware unless optional, encrypted iCloud sync is manually enabled.

Frequently Asked Questions (FAQ)

1. Does Valin require an internet connection to search and summarize saved content? No. Once the ~3.43 GB AI model files (Gemma 4, Whisper, and EmbeddingGemma) and your saved content are downloaded onto your iOS or iPadOS device, features like Ask, Search, offline reading, recipe extraction, and summaries work entirely offline. An internet connection is only required to retrieve new source content from web apps, download the initial models, sync via iCloud, or open original website links.
2. How much device storage and RAM does Valin need to run? Valin requires a compatible iPhone or iPad running iOS 26 or iPadOS 26 or later with at least 8 GB of RAM (such as iPhone 15 Pro models and newer). In terms of storage, you need approximately 3.43 GB of free space dedicated to the on-device AI model files, plus additional headroom to accommodate app caches, system overhead, and your saved library content.
3. Are my saved videos and notes sent to a cloud server or third-party AI? No. All AI processing—including video frame analysis, audio transcription via Whisper, and semantic embedding generation—happens locally on your device's hardware. There is no Valin account required, and your data remains strictly local unless you choose to enable private, optional iCloud synchronization between your own Apple devices.