Sorted

Product Introduction

1. Definition: Sorted is a local-first, AI-powered file organization and document intelligence application for macOS. It operates as a private, on-device library that ingests, indexes, and manages files using local machine learning models, optical character recognition (OCR), and rule-based automation.
2. Core Value Proposition: It exists to solve digital clutter and information retrieval challenges by transforming a disorganized Mac file system into a perpetually self-organizing, semantically searchable knowledge base. Its primary value is private, automated file organization with AI-powered search and document Q&A, all processed locally without cloud dependency.

Main Features

1. Local AI Processing & Semantic Search: Sorted uses on-device embedding models to create vector representations of document content (text extracted via OCR from PDFs, images, and Office files). This enables hybrid search that combines traditional keyword matching with semantic search by meaning, allowing users to find files based on conceptual queries rather than exact filenames.
2. Document Q&A with Citations: The "Ask Sorted" feature employs a local, lightweight Large Language Model (LLM) to answer natural language questions based on indexed document content. Every answer includes specific citations linking directly to the source file and page (using PDFKit for PDFs), ensuring verifiability and trust. It supports follow-up conversations and voice input via local speech recognition.
3. Automated Organization with Custom Rules & Confidence Controls: Users can define custom organization rules based on file attributes, extracted content, or categories. These rules (which always override AI suggestions) can move files and add tags. The system provides a confidence-based approval queue, allowing high-confidence AI suggestions to run automatically while holding uncertain ones for manual review, balancing automation with user control.
4. Watched Folders & Continuous Processing: Sorted can monitor designated folders (like Downloads or Scans) for new or changed files. It recursively copies new items into its private library for processing—including duplicate detection, text extraction, indexing, and rule application—while leaving the original source files untouched, enabling ongoing organization without workflow disruption.
5. Exact & Visual Duplicate Detection: The tool performs byte-for-byte duplicate detection and visual similarity analysis for images and scans. It presents duplicate candidates in a review interface, showing file paths and context, allowing users to make informed decisions about what to keep or delete, addressing both renamed copies and visually identical files.
6. Tax Pack Export: A specialized workflow for financial document management. Sorted extracts transaction data from statements, invoices, and receipts, allowing users to query totals, merchant summaries, and monthly spending. Users can then export a reviewable Tax Pack, which includes original documents, CSV summaries of documents and transactions, file hashes, and an activity log for accountant-ready packages.

Problems Solved

1. Pain Point: Chronic digital disorganization where important documents (receipts, statements, contracts, scans) accumulate in Downloads and desktop folders with unhelpful names (e.g., scan_0042.pdf), making them effectively lost and creating anxiety around tax time or client work.
2. Target Audience: Mac-using professionals and individuals who handle diverse document types, including freelancers, consultants, small business owners, researchers, students, and anyone burdened by personal document management. Specific personas include individuals with ADHD who struggle with executive function tasks like filing, and professionals like lawyers, doctors, and accountants who need reliable retrieval of client records.
3. Use Cases:
   • Tax Preparation: Automatically gathering, categorizing, and querying financial documents from across a system, then exporting an auditable package.
   • Client/Project File Management: Using custom rules to automatically route contracts, invoices, and deliverables to correct client folders based on content.
   • Personal Knowledge Base: Creating a searchable, private archive of medical records, insurance policies, and vehicle documents where information can be retrieved via natural language questions.
   • Inbox Zero for Downloads: Using watched folders to ensure new downloads are processed, deduplicated, and organized immediately without manual intervention.

Unique Advantages

1. Strengths & Limitations (Pros & Cons):

   • Pros:
     • Uncompromising Privacy: All processing (OCR, embedding, AI Q&A) occurs on-device; no file contents or queries are uploaded to external servers.
     • High Verifiability: Cited answers and a transparent activity history allow users to audit every automated action and AI-generated response.
     • One-Time Purchase Model: Priced at a single $9.99 fee, it offers a compelling alternative to subscription-based cloud services.
     • Deep macOS Integration: Uses native frameworks like Vision for OCR, PDFKit, and Quick Look, ensuring performance and system compatibility.
   • Cons:
     • Platform Lock-in: Exclusively available for Apple silicon Macs running macOS 14+, excluding Windows, Linux, and older Mac users.
     • Local Resource Dependency: Performance and model capability are constrained by the user's local hardware (CPU, GPU, RAM).
     • Limited Scope: Primarily a personal/document library tool; not designed for real-time team collaboration or as a full-scale Document Management System (DMS).

2. Key Alternatives & Differentiation:

   • vs. Hazel (Automation Utility): Hazel is a powerful, rule-based folder automation tool. Sorted differentiates by adding a local AI layer for content understanding and semantic search, which Hazel lacks. Hazel excels at surface-level file actions (rename, move based on metadata); Sorted understands file content for organization and Q&A.
   • vs. Devonthink (Document Database): Both are local-first, search-heavy Mac applications. Devonthink is a more powerful, complex database suited for heavy research. Sorted is more user-friendly and opinionated towards automatic organization, with a stronger focus on set-and-forget watched folders and a simpler UI for everyday document chaos.
   • vs. Cloud Services (Google Drive, Dropbox with AI): Cloud services offer AI features but require uploading data to their servers. Sorted's core differentiation is complete on-device privacy. It is for users who prioritize data sovereignty over the potentially more powerful, cloud-based AI models offered by large tech companies.

Frequently Asked Questions (FAQ)

1. Is Sorted AI file organizer secure and private? Yes, Sorted is designed for maximum privacy. All AI processing, including document text extraction, semantic indexing, and question answering, is performed locally on your Mac. Your file contents and search queries are never uploaded to any external server.
2. How does Sorted's AI file organization work without the internet? Sorted uses compact, pre-downloaded machine learning models that run directly on your Mac's hardware (Apple silicon or Intel). It performs Optical Character Recognition (OCR) using Apple's Vision framework and uses local embedding models for semantic search, requiring no internet connection for core organization and Q&A features.
3. Can Sorted automatically organize my Downloads folder? Yes, through its Watched Folders feature. You can set your Downloads folder as a watched source. Sorted will automatically detect new files, copy them into its private library, and process them with your custom rules and AI suggestions, keeping the original files in place.
4. What file types does Sorted support for content search? Sorted extracts and indexes text from a wide range of formats: PDFs, images (JPEG, PNG, etc.) using OCR, Microsoft Office documents (Word, Excel, PowerPoint), as well as plain text, CSV, Markdown, JSON, RTF, and source code files. Unsupported formats remain searchable by filename and tags.
5. What is the difference between Sorted and a traditional file search tool like Spotlight? While Spotlight indexes file metadata and content, Sorted adds a semantic understanding layer. You can search by concept (e.g., "insurance renewal document") not just keywords. Furthermore, Sorted provides automated organization with rules, duplicate detection, document Q&A with citations, and specialized workflows like Tax Pack export, which are beyond the scope of a system search tool.