Impact Forge

Building for a Better World

Comment

ImpactForge: Building for a Better World

MVP: SightCom – Unlocking Vision Through AI Communication

Inspiration

ImpactForge was created to address real-world challenges with innovative, inclusive technology. SightCom, our first MVP, is inspired by conversations with visually impaired individuals who struggle with daily information access. Witnessing how technological barriers limit independence and opportunity, we set out to build solutions that restore agency and dignity to all. The core drive: to empower people, regardless of ability, with the tools they need for a better life.

What it does

SightCom is an AI-powered web tool that translates images and text into spoken descriptions—instantly, simply, and accessibly.

  • Users upload or capture an image.
  • The app extracts any visible text (street signs, menus, forms) via AI OCR.
  • SightCom describes the scene and reads the text aloud, enabling visually impaired users to access critical information on-the-go.
  • The interface is designed to be fully accessible: high-contrast UI, large buttons, keyboard navigation, and clear audio feedback.
  • MVP demo also allows for instant feedback submissions, powering our build-in-public loop.

How we built it

  • User Flow: We designed a streamlined, one-click upload-to-speech system focusing on accessibility first.
  • UI/UX: Figma prototype for easy iteration and validation with real users.
  • Frontend: React (JavaScript) for the web interface.
  • Backend/API: Used Python (for prototyping OCR/TTS flow where needed), integrated Google Cloud Vision API for robust text extraction.
  • Speech Output: Web Speech API delivers real-time text-to-speech descriptions.
  • Rapid Iteration: Hosted on Replit/Vercel, continuous feedback loop via Instagram, Reddit, and Discord communities.

Challenges we ran into

  • Accessibility: Ensuring a seamless, accessible experience across devices—including for users who rely on screen readers or keyboard navigation—required deep user involvement and testing.
  • AI Interpretation: Some open-source and even cloud APIs struggled with unusual photos, handwritten text, or poor lighting conditions.
  • Data Privacy: Processing sensitive images with trust and transparency was crucial.
  • Time Constraint: Validating with real visually impaired users and rapidly iterating major UI changes in just one week was demanding but rewarding.

Accomplishments that we're proud of

  • MVP Launch: Delivered a fully functional SightCom MVP, validated with real user feedback in less than a week.
  • Community Impact: Early posts and demos gained traction on Instagram and Reddit accessibility groups, resulting in authentic stories and feedback from visually impaired users.
  • Design Excellence: High-contrast, accessible-first design praised by testers for its simplicity.
  • Built in Public: Shared daily progress and feedback, with several users engaging, suggesting features, and expressing interest in future use.

What we learned

  • Direct User Involvement is Essential: Solutions for accessibility only succeed when built in close partnership with those who face the challenges every day.
  • Speed & Flexibility Wins: Quick iteration, public sharing, and the willingness to pivot based on user feedback makes a better product and builds trust.
  • Impact Over Perfection: A scrappy, well-targeted MVP can create real value and open doors for partnerships—even without all bells and whistles.

What's next for ImpactForge

  • Expand SightCom: Add support for non-Latin scripts, offline mode, and on-device processing for greater privacy and reach.
  • Mobile App: Develop native iOS/Android versions for broader accessibility.
  • Partnerships: Collaborate with schools, NGOs, and public agencies to deploy SightCom to those who need it most.
  • New MVPs: Stay true to our mission—using ImpactForge as a launchpad for more socially responsible tech products in energy, health, and civic good.

Built with

  • Languages: JavaScript (React), Python (backend)
  • Frameworks: React, Node.js (optional), Flask/FastAPI (optional)
  • APIs: Google Cloud Vision (OCR), Web Speech API (TTS), META llama2 (for advanced descriptions, if available)
  • Platforms: Replit, Vercel/Netlify, GitHub
  • Design: Figma, Canva
  • Social/Validation: Instagram, Reddit, Discord

Built With

Share this project:

Updates