FEEDO

Feedo is a voice-first feedback tool that lets users share insights via audio or forms. Track feedback, analyze it with AI, and prioritize features—all from one embeddable widget.

Comment

Inspiration

I wanted a better way to hear what users really think. Traditional forms felt lifeless, and written responses often missed the tone, urgency, or emotion behind the feedback. I realized that voice is the most natural way we communicate—so why isn’t it used more in feedback collection? That’s what inspired me to build Feedo, a voice-first feedback tool that makes giving feedback as easy as leaving a voice note—then turning that voice into actionable insights using AI.

What it does

Feedo is a voice-first feedback platform that lets users record short audio messages directly from an embeddable widget. It also supports form-based input, but the core focus is to capture rich, emotional feedback through voice. Each voice message is automatically transcribed using ElevenLabs, then analyzed with OpenAI to generate insights, detect sentiment, and group similar feedback. All responses are linked to targeted campaigns and visualized in a clean dashboard, helping teams quickly understand what users are saying and why it matters.

How I built it

Feedo is made up of two parts: a Next.js main app and a React + Vite embeddable widget, all powered by modern serverless infrastructure and AI integrations.

  • Main App (Next.js + Supabase)

    • Built with Next.js and deployed via Vercel.
    • Handles the dashboard UI, campaign setup, roadmap, and API routes.
    • Uses Supabase for:
    • Authentication
    • Database (PostgreSQL)
    • Edge Functions and Row-Level Security for fine-grained data control.

  • Widget App (React + Vite)

    • A standalone React widget built with Vite, designed to be embedded via <script> on any website.
    • Uses the MediaRecorder API to capture voice input in-browser.
    • Deployed using Wrangler to Cloudflare Pages, with CDN distribution for ultra-fast loading.

  • Voice Transcription and AI Insights

    • Audio recordings are transcribed using ElevenLabs, providing fast and accurate speech-to-text.
    • Transcriptions are then analyzed using OpenAI, which generates insights, detects sentiment, and clusters similar feedback.

This modular architecture allows Feedo to remain fast, scalable, and smart—offering voice-first feedback collection with AI-powered understanding built right in.

Challenges I ran into

To build the embeddable widget, I had to separate it into its own standalone project using React and Vite. That introduced challenges with deployment, script injection, and making sure it stayed lightweight yet functional. I also ran into several issues around audio transcription and sentiment analysis—from handling microphone permissions to managing audio uploads, then reliably transcribing and analyzing them using external APIs like ElevenLabs and OpenAI.

Another challenge was designing a smooth and intuitive voice-first UX that didn’t feel clunky or overwhelming. Since users might be recording in different contexts, I had to strike the right balance between simplicity, customization, and responsiveness. As a solo builder, managing both the tech and the user experience took a lot of iteration.

Accomplishments that I'm proud of

  • Built and deployed the entire product solo—from the backend to the AI pipeline to the frontend UI.
  • Created a fully embeddable voice-first widget, deployable via <script> with no iframe needed.
  • Integrated ElevenLabs for accurate speech-to-text transcription of voice feedback.
  • Connected OpenAI to automatically generate insights, sentiment, and categorize feedback.
  • Designed a clean, intuitive dashboard for managing campaigns, roadmaps, and user input.
  • Deployed the main app via Vercel and the widget via Cloudflare CDN, ensuring performance and scalability.
  • Launched everything under my own domain: getfeedo.com.
  • Populated the app with realistic demo data and audio samples to simulate live usage.

What I learned

  • Being solo means wearing every hat — from designing the user experience to setting up CDN deployments and handling auth flows, I had to do it all. It forced me to prioritize what truly matters to users.
  • Embeddable widgets need to be frictionless — performance, script loading, and UI simplicity all matter when you're inserting a widget into someone else’s site.

What's next for FEEDO

  • Add payment integration for subscription plans and billing packages
  • Launch email + voice ai reports, auto-generated audio feedback reports
  • Enable feature announcements via the audio widget
  • Expand AI to detect themes and trends across feedback
  • Offer more widget customization and branding options
  • Continuously listen to users and iterate to make Feedo the perfect voice-first feedback tool
  • Adding the widget through npm package and mobile SDK.

Built With

Share this project:

Updates