DeskGraph

The description states that DeskGraph is a local-first computer context graph for folders explicitly chosen by the user. It builds a local SQLite manifest, supports bounded local document extraction, and makes content searchable using SQLite FTS5, without uploading data to the cloud.

It includes:

• Explicit native folder authorization
• Metadata-only manifest scan
• Bounded local extraction of text, Markdown, source code, PDFs, DOCX, PPTX, XLSX
• Offline lexical search for Traditional Chinese and English
• A read-only MCP search_files tool
• Suggestion-only Smart Cleanup Inbox and durable Preview-only cleanup features

The product is described as not uploading filenames, paths, document text, OCR, embeddings, or graph data by default.

Inference The system appears to be a desktop application built with Rust for backend logic, Tauri/React for UI, and SQLite FTS5 for search. It uses Codex (GPT-5.6) as a development collaborator but does not appear to integrate AI inference in its core functionality.

The author states that DeskGraph addresses the narrower, practical problem of giving users and AI agents a controllable local view of explicitly selected context, without asking people to hand over their entire computer or trust opaque automation.

It positions itself as:

• A privacy-first alternative to “AI organizer” tools
• Focused on evidence, preview, and refusal over unsafe automation
• Not allowing file changes unless identity validation, durable transactions, recovery, and Undo are implemented

The claim evolution shows a focus on safety and control, not broad AI-powered automation or cloud integration.

Inference The positioning is clearly focused on local-first privacy, with an emphasis on user control and safety. It does not appear to be a general-purpose AI assistant or productivity tool, but rather a contextual data layer for local file systems.

The description states that DeskGraph targets users who want:

• A controllable local view of explicitly selected context
• To avoid handing over their entire computer or trusting opaque automation
• A system where file changes are not permitted without identity validation, durable transactions, recovery, and Undo

It is described as a tool for people who value privacy and safety in AI interactions with personal files.

No specific customer segments, personas, or use cases are described beyond this general positioning.

The description does not provide any information about:

• Revenue model
• Pricing structure
• Monetization strategy
• Customer acquisition or retention plans

Not evidenced.

The project is built with:

• Rust for filesystem policy, SQLite state, extraction jobs, graph logic, and MCP
• Tauri + React + TypeScript for desktop UI
• SQLite FTS5 for lexical search
• Codex (GPT-5.6) used as a development collaborator

It includes:

• Bounded local document extraction
• Read-only MCP tool
• Preview-only Rename and Cleanup features
• Safety decisions made to prevent unsafe automation (e.g., preview-only rename)

The author notes that this is a pre-release development build, not a v0.1 release.

Inference The technical stack suggests a local-first, privacy-focused desktop application with strong emphasis on safety and control. It is not yet production-ready, as noted by the author.

The description states:

• This is a pre-release development build
• It was submitted to the OpenAI 2026 hackathon
• The project is not a completed v0.1 release
• It has been tested with a synthetic demo folder and verified on macOS arm64
• No production file actions, Trash, recovery, or Undo are enabled

There is no evidence of:

• Customers
• Revenue
• Usage metrics
• Product adoption
• Market traction

Not evidenced.

The description does not mention any competitors or direct market context.

Not evidenced.

• Pre-release status: The product is described as a pre-release development build, not a finished product.
• No commercial traction: No evidence of users, customers, or revenue.
• Limited scope: The system only supports local file scanning and search; no AI inference or automation beyond preview.
• No monetization strategy: No indication of how the project will generate revenue.
• Self-reported only: All claims are unverified self-descriptions.

Inference The project is in a very early stage, with no commercial viability or traction evident. It may be a proof-of-concept or prototype, not a product ready for market.

1. What is the intended path from this pre-release build to a production-ready product?
2. Are there any plans for monetization or customer acquisition beyond personal use?
3. What are the key technical challenges remaining before full functionality (e.g., incremental indexing, Trash, Undo)?
4. How does the author plan to validate user needs and adoption beyond their own development work?
5. Is there any interest from users or partners in using or investing in this project?