DEV Community: Chhavi Gupta

Vision Assistant v2.0

Chhavi Gupta — Wed, 21 Oct 2026 17:43:37 +0000

This is a submission for the Hacktoberfest Open-Source AI Challenge Week 3: Slow and Steady

Vision Assistant v2.0 — Asynchronous Screen Intelligence & Background AI Pipeline

What I Built

Vision Assistant v2.0 is an open-source desktop utility that transforms instant system-wide screen captures into deep, asynchronous AI reasoning pipelines.

Instead of locking up your workflow while waiting for heavy vision models to parse dense screens, codebases, or charts, Vision Assistant captures the context via hotkey and dispatches it into a background execution queue. You can keep working, while Vision Assistant asynchronously breaks down the visual context, runs multi-step AI agent workflows, and delivers structured reports directly to your workspace.

Who gets to be a little lazier?
Developers, agency operators, and researchers who constantly analyze complex UI screens, dashboards, or documentation and don't want to waste time manually copying text or waiting for live API responses.

Demo

GitHub Repository: Vision Assistant Repository
Features Showcase: System-wide keyboard shortcut capture, background async processing queue, multi-agent context analysis.

Code

{https://github.com/chhavigupta783/vission-assisstant-v1}

How I Built It

Desktop System Layer: Built using Electron & C++/Python modules to handle global system hotkeys, screen region cropping, and low-footprint desktop execution.
Async Queue & Automation: Integrated with n8n webhooks and background worker queues to process heavy multi-modal tasks asynchronously without UI lag.
AI Agent Orchestration: Powered by open-weight vision models and AI agent frameworks that decompose captured visual context into actionable text, code snippets, or structured summaries.

Why Does Open Innovation Matter?

Building a system-level vision assistant requires complete control over data privacy, custom local models, and self-hosted automation pipelines. Open innovation allows us to route sensitive desktop screen context through self-hosted open-weight models and custom background workers—something locked proprietary ecosystem APIs cannot deliver with full user privacy.

The Lazy Payoff

By offloading complex screen analysis, log debugging from visual terminals, and automated web layout breakdown to Vision Assistant's background queue, I saved over 5 hours of repetitive context-switching this week. I trigger the capture, keep working, and view the fully parsed breakdown when it's ready.

Prize Categories

Main Challenge: Slow and Steady (Long-running & Asynchronous AI Workflows)
Open-Source AI Advocate

Vision Assistant: Streamlining Client Workflows with Privacy-First Desktop AI

Chhavi Gupta — Sat, 17 Oct 2026 10:52:36 +0000

This is a submission for the Hacktoberfest Open-Source AI Challenge Week 2: Side Hustle

What I Built

I built Vision Assistant v1.0.0 (and my suite of custom B2B n8n AI Automations) designed specifically for freelancers, small business owners, and agency teams who need instant, context-aware AI assistance directly on their desktop without friction or monthly cloud subscriptions.

Vision Assistant sits quietly in the background, activates instantly via Ctrl + Shift + Space, and handles multi-screen desktop captures, local memory tracking, and vision analysis—allowing professionals to automate tedious review tasks, capture field/client notes, and streamline complex workflows in seconds.

Demo

Video Walkthrough & Setup Guide: Google Drive Demo Folder
Executable Package: Download Vision.Assistant-1.0.0-win.zip to run natively on Windows.

Code

(https://github.com/chhavigupta783/vission-assisstant-v1)

How I Built It

Electron & electron-builder: Native desktop application framework packaged cleanly for Windows.
Global Shortcut Listener: Triggers system-wide context overlay with zero browser context-switching.
Desktop Capturer & Web APIs: Captures multi-monitor regions simultaneously for instant LLM vision analysis.
Local Context Engine: Stores user preferences and project sessions locally to keep client data private.

Why Does Open Innovation Matter?

For a business-focused tool, privacy and zero API overhead make or break the business model:

100% Data Privacy: Clients and agency users won't put sensitive financial documents or proprietary code into closed web portals. Local/open processing keeps data on their machine.
Predictable Margins: Closed APIs charge per token/call, making utility tools expensive to scale. Open innovation allows delivering high-value workflow automation at near-zero marginal cost.

Target User / First Traction

Targeted at Digital Marketing Agencies & Freelance Developers who waste 10+ hours a week switching between tabs and manually analyzing screenshots or lead forms. The feedback so far confirms that having a zero-latency, system-wide AI assistant saves significant operational time daily.

Prize Categories

Best Use of Render (if hosted on Render)
Best Use of Gemma (if integrated with Gemma/open weights)
Best Use of a .tech Domain

Built by Chhavi Gupta

Vision Assistant v1.0.0: A System-Wide Desktop AI Tool Built for Fast Workflows

Chhavi Gupta — Thu, 08 Oct 2026 18:58:59 +0000

This is a submission for the Hacktoberfest Open-Source AI Challenge Week 1: Touch Grass

What I Built

Meet Vision Assistant v1.0.0, a custom desktop AI vision tool built from scratch using Electron, designed to bring instant, system-wide AI assistance directly to your desktop environment without the friction of constant context-switching.

While the "Touch Grass" challenge pushes us to get away from endless screens, outdoor explorers, hikers, botanists, and field researchers often need to quickly process field notes, identify trail maps, capture outdoor specimens, or organize trip logs without being chained to heavy web browsers or clunky cloud tools. Vision Assistant bridges this gap: it sits quietly in the background, acts instantly via global shortcuts, and captures multi-monitor or desktop display areas to give you immediate, contextual AI insights so you can look up what you need and get right back outside into the world.

Demo

Setup & Walkthrough Guide: Watch the full setup and demonstration video on Google Drive Setup Guide[cite: 1].
Installation Package: Download the latest standalone release package ((https://github.com/chhavigupta783/vission-assisstant-v1/releases/download/v1.0.0/Vision.Assistant-1.0.0-win.zip) ) from the releases page to run it locally on Windows.

Code

The project is architected as a native desktop application wrapper utilizing Electron, web capture APIs, and custom IPC handlers.
(https://github.com/chhavigupta783/vission-assisstant-v1)

How I Built It

Vision Assistant is built around open-source architecture principles:

Electron Framework: Powers the native desktop application wrapper, packaged seamlessly via electron-builder into a standalone Windows executable.
Global Shortcut Manager: Listens discreetly in the background for the Ctrl + Shift + Space key combination, letting you trigger the assistant instantly from any active window or application.
Multiple Screen Capture: Integrates native desktop screen-capturing capabilities to grab multiple screens or display areas simultaneously, perfect for complex field workflows and multi-monitor setups.
Local Context & Memory: Tracks ongoing development and user project preferences across sessions to keep responses contextual.

Why Does Open Innovation Matter?

Building Vision Assistant using open-source frameworks (Electron and local/open-weight processing capability) matters because privacy and reliability are paramount. When analyzing sensitive desktop screens, documents, or personal field data, relying on rigid closed APIs introduces security risks and unwanted cloud dependencies. An open architecture gives developers full control over system-level bindings, custom keyboard hooks, local context memory, and offline execution—ensuring total data sovereignty and zero dependency on closed-source constraints.

Prize Categories

Best Use of Open-Source AI / Local Inference
Best Desktop / Electron Tooling

Building Vision Assistant: A Custom Desktop AI Tool Built with Electron

Chhavi Gupta — Sun, 27 Sep 2026 10:35:25 +0000

This is a submission for the MLH x DEV Writing Challenge

What I Built

I built Vision Assistant v1.0.0, a custom desktop AI vision tool created from scratch using Electron.

What it does: It brings instant, system-wide AI assistance directly to your desktop environment without the friction of constant context-switching. It features global shortcut activation (Ctrl + Shift + Space), multi-screen capture capabilities, real-time chat & vision analysis, and memory persistence across sessions.
The Inspiration: I wanted to eliminate the workflow friction of switching between browser tabs and separate tools when analyzing multiple screen areas or seeking quick AI assistance during development.
What I Learned: Building a desktop app from scratch using Electron taught me a lot about handling system-level global shortcuts, managing multi-screen capture streams, and packaging a clean Windows executable using electron-builder.

Demo

You can check out the setup guide video and walkthrough here:
Vision Assistant Setup Guide Video

You can also download the latest Windows standalone release package directly from the GitHub repository:
GitHub Repository & Releases
release link for windows user

Partner Technologies

For this project, I focused heavily on core web and desktop ecosystems using JavaScript (69%) and HTML (31%), powered by Electron to bundle a seamless standalone Windows application with a dark/purple customized UI theme optimized for desktop environments.

Hackathon Experience

Building Vision Assistant for this challenge was an incredible journey of pushing desktop development boundaries. Bringing an idea from local scripts to a fully packaged, global shortcut-enabled desktop application made it a deeply rewarding learning experience!

Say Hello to Vision Assistant: Your Instant, Shortcut-Triggered Desktop AI Tool

Chhavi Gupta — Tue, 22 Sep 2026 18:28:49 +0000

``This is a submission for the Sanity Challenge, Path Two: Vibe-Code Something Strange

What I Built

I built Vision Assistant v1.0.0, a custom desktop AI vision tool built from scratch using Electron. It is designed to bring instant, system-wide AI assistance directly to the desktop environment without the friction of constant context-switching and this version package is built only for windows user right now.

Key Features:

Global Shortcut Activation:
Runs in the background and triggers instantly from any active window or application using the Ctrl + Shift + Space shortcut.

Real-Time Chat & Vision:
Seamlessly process and analyze visual content using integrated AI capabilities right from the desktop.

Multiple Screen Capture:
Allows capturing and analyzing multiple screens or active display areas simultaneously, perfect for multi-monitor workflows.

Native Windows Experience:

Lightweight standalone Windows application packaged via electron-builder with a sleek dark/purple UI theme. )

Demo

Setup Guide & Walkthrough Video:
[https://drive.google.com/drive/folders/1Kci00sPzQalZA78Urge4v8mkXLHp4dbR?usp=sharing]

Explore the repository and release notes on GitHub:
GitHub Release Page

Grab the direct package:
Direct Download Link

You can download the latest Windows release bundle directly from the GitHub repository releases.

Code

GitHub Repository:

https://github.com/chhavigupta783/vission-assisstant-v1

My Build Process

AI-Native Setup:
Leveraged AI capabilities to vibe-code and structure the entire Electron desktop app from scratch, handling native window bindings and global shortcuts.

AI Integration:
Powered by the Gemini API to handle real-time chat, vision analysis, and contextual interactions seamlessly.

Challenges Faced:
Managing local environment configurations, ensuring smooth packaging with electron-builder, and handling window focus events for system-wide shortcuts.

Course-Correction:

Refined the build setup documentation to guide users properly through extracting zip assets and bypassing Windows SmartScreen warnings for custom unsigned executables.

Sanity Project Details

Note: While the core application is built natively using Electron and the Gemini API, Sanity's documentation was referenced for content modeling patterns and architecture guidelines.

[A sleek, professional 3D graphic featuring a vibrant purple glowing bubble that showcases the dark-mode UI of the Vision Assistant desktop application. At the bottom, a stylish UI element clearly highlights the core system-wide feature with the text "press ctrl + shift + space".]

(https://dev-to-uploads.s3.us-east-2.amazonaws.com/uploads/articles/0ja24tx9sglq2sco303q.png)

<!-- Thank you for visiting --!>

DEV Community: Chhavi Gupta

Vision Assistant v2.0

Vision Assistant v2.0 — Asynchronous Screen Intelligence & Background AI Pipeline

What I Built

Demo

Code

How I Built It

Why Does Open Innovation Matter?

The Lazy Payoff

Prize Categories

Vision Assistant: Streamlining Client Workflows with Privacy-First Desktop AI

What I Built

Demo

Code

How I Built It

Why Does Open Innovation Matter?

Target User / First Traction

Prize Categories

Vision Assistant v1.0.0: A System-Wide Desktop AI Tool Built for Fast Workflows

What I Built

Demo

Code

How I Built It

Why Does Open Innovation Matter?

Prize Categories

**Building Vision Assistant: A Custom Desktop AI Tool Built with Electron**

What I Built

Demo

Partner Technologies

Hackathon Experience

Say Hello to Vision Assistant: Your Instant, Shortcut-Triggered Desktop AI Tool

``This is a submission for the Sanity Challenge, Path Two: Vibe-Code Something Strange

What I Built

** Key Features: **

Lightweight standalone Windows application packaged via electron-builder with a sleek dark/purple UI theme. )

Demo

You can download the latest Windows release bundle directly from the GitHub repository releases.

Code

https://github.com/chhavigupta783/vission-assisstant-v1

My Build Process

Refined the build setup documentation to guide users properly through extracting zip assets and bypassing Windows SmartScreen warnings for custom unsigned executables.

Sanity Project Details

Note: While the core application is built natively using Electron and the Gemini API, Sanity's documentation was referenced for content modeling patterns and architecture guidelines.

Building Vision Assistant: A Custom Desktop AI Tool Built with Electron

Key Features: