All projects

Case study · omnispot

Desktop search, navigation, and file actions

OmniSpot

A fast command-space for local Windows search, folder navigation, previews, and file actions.

Classic search runs entirely on-device. AI mode sends the query text to Groq when connected and falls back to a narrower local parser when offline.

  • .NET 8
  • WPF
  • Windows 10/11
  • 72 automated tests
  • Active development
View source on GitHub
  • Desktop
  • AI
OmniSpot main folder view using a synthetic DemoWorkspace
Role
Solo developer
Team
Independent build
Timeframe
2025 - 2026
Status
Active iteration
Platform
Windows desktop utility
Core file search stays local; Groq is an explicit, optional step used only to parse AI-mode queries.
01

The project at a glance

Context

An independently developed Windows desktop utility and an ongoing study in faster, keyboard-first file interaction.

Project overview

OmniSpot is a lightweight desktop file explorer built for fast everyday access on Windows. Classic filename search, indexing, folder navigation, previews, and file actions run locally. AI mode sends only the query text to Groq for intent parsing when connected; without a connection, it falls back to a narrower local parser. The app supports multiple indexed directories, inline copy/cut/paste/rename/delete actions, and image or video thumbnails without invasive system changes.

02

Why this exists

Problem & direction

Problem

Traditional file navigation on Windows is still too slow for high-frequency tasks. Reaching the right directory, previewing content, and performing small actions often means bouncing between Explorer windows, context menus, and search boxes. I wanted a single interface that could be opened instantly and stay local, fast, and respectful of the system.

Goal

Create a launcher-like workspace where search, folder traversal, previews, and file actions feel part of one coherent system instead of separate tools.

03

How it comes together

Experience flow

  1. 01

    Open instantly

    A global hotkey makes the interface feel like a utility layer rather than a separate app window.

  2. 02

    Search with intent

    Classic filename search stays local. AI mode sends query text to Groq when online and uses a more limited local parser when offline.

  3. 03

    Confirm visually

    Thumbnail previews reduce mistakes before open, rename, move, or delete actions.

  4. 04

    Act without leaving context

    Common file operations stay inside the same surface, which keeps the flow fast and focused.

04

See the work

Evidence & media

Available now

Interface walkthrough

Six privacy-safe screens document the main view, local search, Groq-assisted intent search, folder previews, file actions, and index settings.

Interface gallery

OmniSpot synthetic main folder view
Main folder view built from a synthetic DemoWorkspace with no personal paths or filenames.
OmniSpot local filename search results
Classic indexed filename search running locally on the Windows device.
OmniSpot Groq-assisted natural-language search
Natural-language search with Groq intent parsing and an explicit limited offline fallback.
OmniSpot folder navigation and thumbnail previews
Folder traversal with breadcrumbs, synthetic folders, and small visual previews.
OmniSpot right-click file operations menu
Right-click file actions for open, copy, cut, rename, and delete without leaving the launcher.
OmniSpot settings and index status
Settings view showing synthetic indexed locations, item count, and watcher status.
05

What it is made of

Technical shape

Feature set
  • Global hotkey access with Ctrl+Space from anywhere in Windows
  • Fully local indexed filename search
  • Optional Groq intent parsing for AI mode with a limited local fallback when offline
  • Folder navigation directly inside the launcher interface
  • Built-in copy, cut, paste, delete, and rename workflows
  • Thumbnail previews for image and video files
  • Color-coded folder visualization for faster scanning
  • Multi-directory indexing for broader workspace coverage
  • 72/72 automated tests covering search, concurrency, indexing, watcher, navigation, settings, and connectivity behavior
Architecture

Structured as a .NET 8 and WPF solution with a separate core layer for filesystem models, indexing, tokenization, scoring, watching, and scanning. The latest release build completed with 0 errors and 2 known NU1701 warnings. All 72 automated tests passed across search, concurrency, indexing, watcher, navigation, settings, and connectivity behavior.

Stack
.NET 8C# 12WPFWindows desktop APIscustom indexing and search structures
06

Engineering notes

Decisions & challenges

Balancing speed with context

Challenge Search launchers are often fast because they hide detail. File explorers are detailed because they slow you down with extra structure.

Response OmniSpot tries to keep just enough hierarchy visible to preserve confidence while still behaving like a fast command surface.

Making natural-language search useful

Challenge Natural language sounds impressive, but poor relevance quickly destroys trust in a local utility.

Response The boundary is explicit: Groq receives the query text only when AI mode is used online, while local index matching stays on-device. Offline, a narrower local intent parser takes over.

Keeping file actions safe

Challenge Speed is only valuable if the user feels in control before copy, move, rename, or delete operations.

Response Preview surfaces, folder cues, and scoped operations were designed to reduce destructive mistakes and ambiguity.

07

Reflection

What I learned

This project strengthened my desktop engineering skills around WPF architecture, global hotkey handling, safe local file operations, and designing a natural-language search experience on top of indexed filesystem data.

08

Where it can go

Next steps

  1. 01

    Push the natural-language layer further with better query understanding.

  2. 02

    Expand richer preview states for more file types and larger workspaces.

  3. 03

    Refine ranking, keyboard flow, and packaging for broader everyday use.