Dotient

Product Introduction

1. Definition: Dotient is a local-first desktop application that functions as a privacy-centric file organizer and visual search engine. It leverages on-device machine learning (ML) to create searchable embeddings from your personal files.
2. Core Value Proposition: Dotient exists to solve the problem of finding specific files in large, disorganized digital archives without relying on cloud services, manual tagging, or remembering exact filenames. Its core proposition is enabling private, offline, ML-powered visual search for your personal documents, photos, and videos.

Main Features

1. ML-Powered Visual Search: This is the core technical feature. Dotient uses a quantized vision model and a text model to analyze imported files and create unique digital signatures or "embeddings." Users perform searches using natural language descriptions of the file's content (e.g., "beach sunset," "team meeting in office") instead of file names or manual tags. The system returns results based on visual and semantic similarity. "Signal Tuning" allows users to further refine this search by selecting correct results, effectively creating personalized search filters to improve future accuracy.
2. Local-First Architecture & Privacy: The application operates entirely on the user's machine. During the "Import Your Files" step, files are processed locally; the raw data and their generated ML embeddings never leave the device. This ensures complete data privacy and offline functionality, eliminating dependency on internet connectivity and cloud storage subscriptions.
3. Graph View Visualization: Dotient provides a node-based graph view to visualize the entire file archive. Files are represented as nodes, and edges are dynamically formed between files based on their similarity scores generated by the ML model. This feature automatically detects clusters of related content, allowing users to visually explore their archive, identify patterns, and manually label groups.
4. Cubbies for Organization: Cubbies are flexible digital collections for grouping related files. Users can create Cubbies for projects, mood boards, or specific themes (e.g., "Dad's Birthday"). This feature supports file annotation and acts as a curated layer on top of the ML-powered search, enabling both algorithmic and manual organization.
5. One-Time Purchase & Universal License: Dotient is offered as a one-time purchase (Standard license: $5) for lifetime access. The license covers installation on up to 3 computers and includes all features, offline functionality, future updates, and priority support, contrasting sharply with recurring subscription models common in SaaS file management tools.

Problems Solved

1. Pain Point: Dotient directly addresses the inefficiency and frustration of managing and retrieving files from large, poorly organized personal digital libraries—particularly for visual media where filenames are often generic or non-descriptive. It solves the problem of "I know what it looks like, but I can't find it."
2. Target Audience: The primary users include Photographers & Videographers managing vast shoots; Designers & Art Directors needing quick access to reference images and inspiration; and Individuals with extensive personal photo and file collections lacking systematic organization.
3. Use Cases: Essential scenarios include a photographer instantly locating a specific shot from a multi-day event by describing it; a designer retrieving a mood board for a project; a family user finding a collection of photos of a specific person from years of uploads; or any professional needing to search an archive without relying on pre-existing folder structures or metadata.

Unique Advantages

1. Differentiation: Unlike traditional file managers (Finder, File Explorer) or cloud-based search services (Google Drive), Dotient combines the privacy of local storage with the advanced search capability of computer vision AI. It requires no manual tagging, no cloud upload of sensitive data, and works without an internet connection.
2. Key Innovation: The key innovation is the application of a quantized on-device vision model to enable powerful, descriptive search directly on personal files. Quantization allows the complex ML model to run efficiently on standard consumer desktop hardware. The integration of this search with the similarity-based Graph View and customizable Signal Tuning creates a uniquely interactive and user-tunable file exploration system.

Frequently Asked Questions (FAQ)

1. How does Dotient's visual search technology work? Dotient uses a quantized vision and text model to analyze your files and create mathematical embeddings—digital representations of their content. When you search using a phrase like "red car," your query is converted into an embedding, and the system finds files whose visual and semantic embeddings are most similar to it, all processed locally on your computer.
2. Is my data really private and secure with Dotient? Yes, Dotient's local-first design ensures your files never leave your machine. All processing, embedding creation, and search indexing happen entirely on your device, guaranteeing complete privacy and offline functionality. No data is sent to external servers.
3. Can Dotient replace my current cloud storage and photo apps? Dotient is designed to complement, not necessarily replace, cloud storage. It excels as a specialized, private search and organization layer for files that are stored locally or backed up elsewhere. It's ideal for creating an intelligent, searchable master index of your most important or voluminous personal file archives.
4. What types of files can Dotient search? Based on its vision model technology, Dotient is optimized for visual content. This includes common image formats (JPEG, PNG, etc.), video files (where it can analyze frames), and potentially documents with visual elements (like PDFs or design files). The search works by describing the visual content, not the text within documents.
5. How is the "Signal Tuning" feature different from regular search? Regular search finds matches based on the initial ML model's analysis. Signal Tuning allows you to teach the system your personal preferences. By selecting files that are relevant results for a search term and giving them a label, you create a custom signal that biases future searches for that term toward your preferred style or type of content, making the search increasingly personalized and accurate over time.