Elevate

Elevate is a voice-first AI companion designed to support students through reflection, emotional awareness, and personalized guidance. The product blends cloned voices, AI chat, sharable agent skills, mood tracking, and light gamification into a single experience.

This repository contains the frontend for Elevate. The backend lives in a separate repository and is consumed here through API proxy routes. The frontend is intended to be deployed on Vercel.

What It Does

Elevate helps students feel more supported in their day-to-day lives by making AI feel more personal and emotionally aware.

Core experiences include:

Voice Library: upload or record reference audio to create and manage cloned voices
Personal AI Chat: chat with an assistant using text or voice input
Speech-to-Text: transcribe spoken prompts before sending them into the chat flow
Agent Skills: upload markdown-based skills and browse reusable prompts/instructions
Mood Meter: visualize emotional patterns through a weekly heatmap
Leaderboard: add a motivational layer through positive competition
Profile and Integrations: user profile space with future wearable integration concepts

Why We Built It

Students are often overwhelmed, isolated, and pressured to keep performing even when they are mentally drained. We wanted to explore what a more human-feeling AI experience could look like: one that feels familiar, supportive, and tailored to the student rather than generic.

Elevate is our answer to that idea:

a more personal AI through cloned or familiar voices
a more expressive interface through voice input
a more reflective experience through mood tracking
a more collaborative experience through sharable agent skills

How It Works

The app is built as a Next.js frontend with authenticated product flows and server-side API proxy routes.

High-level flow:

Users sign up or sign in.
Users can upload or record audio to create cloned voices.
In chat, users send text or voice messages.
Voice input is transcribed and forwarded to the backend.
The backend handles core AI and data logic.
The frontend presents additional experiences like agent skills, mood insights, leaderboard views, and profile management.

Architecture

This repo is frontend-only.

Frontend: this repository, built with Next.js and deployed on Vercel
Backend: separate repository/service, not included here
Speech-to-text: integrated through Sarvam AI
Proxy routes: Next.js route handlers forward requests from the frontend to the backend

Examples of proxied capabilities:

authentication
chat requests
cloned voice management
user skills
mood data retrieval

Tech Stack

Next.js 16 with App Router
React 19
TypeScript
Tailwind CSS v4
React Markdown
Sarvam AI speech-to-text integration

Main Product Areas

Route	Purpose
`/home`	Product story and feature journey
`/auth`	Sign in and sign up
`/voices`	Create and manage cloned voices
`/chat`	Chat with the personalized AI assistant
`/agent-skills`	Upload and preview markdown-based agent skills
`/mood-meter`	View mood heatmap and mood meter reference
`/leaderboard`	View rankings and gamified progress
`/profile`	Account view and future integration space

Running Locally

Prerequisites

Node.js 20+ recommended
npm
access to the separate backend service
a Sarvam AI API key for transcription

Environment Variables

Create a .env.local file with:

NEXT_PUBLIC_BACKEND_URL=<your-backend-base-url>
SARVAM_API_KEY=<your-sarvam-api-key>

Install and Start

npm install
npm run dev

Then open http://localhost:3000.

Deployment

The frontend is designed to be deployed on Vercel.

To deploy successfully, configure these environment variables in Vercel:

NEXT_PUBLIC_BACKEND_URL
SARVAM_API_KEY

Because the backend is hosted separately, this frontend depends on that service being reachable from the deployed Vercel app.

Project Structure

app/
  auth/                 # Authentication flow
  home/                 # Product story / landing experience
  voices/               # Voice cloning UI
  chat/                 # Chat experience
  agent-skills/         # Skill upload and preview UI
  mood-meter/           # Mood data visualizations
  leaderboard/          # Gamified rankings
  profile/              # User profile
  api/                  # Next.js proxy routes to backend services
components/             # Shared UI building blocks
context/                # Auth and chat state
hooks/                  # Client-side helpers and behavior
lib/                    # Types, config, utilities, mock data

Current State

Elevate is currently a hackathon-stage product and prototype. Some areas already connect to backend services, while others still use sample or mock data to demonstrate the intended experience.

That makes this repository useful both as: