⚡ Sona AI — Intelligent Multi-Persona AI Workspace

Transforming information into deep understanding, multi-modal synthesis, and real-time intelligence.
Powered by Groq LPU™ ultra-low latency inference, AssemblyAI, Next.js 14 (App Router), and Supabase.

---

👥 Project Leadership & Team

• Project Owner / Lead: Christian01-del
• AI & Backend Developer: reactjay

---

🌟 Overview & Vision

Sona AI is an advanced, persona-driven artificial intelligence platform designed to revolutionize how students, researchers, business professionals, and creators interact with high-density documents, multimedia, and complex knowledge.

Unlike traditional single-purpose chat interfaces, Sona AI unifies ultra-fast LPU inference, split-screen document intelligence with interactive citations, real-time conversational voice calls, and enterprise-grade administrative controls into a cohesive, high-performance workspace.

---

🚀 Key Features

1. 🧠 Multi-Persona Cognitive Modes

Tailor the AI's reasoning depth, tone, and contextual tools based on your specific task:

• Student Mode: Socratic explanations, flashcards, breakdown of formulas, and study guides.
• Business Mode: Executive summaries, actionable memos, strategic SWOT analysis, and ROI metrics.
• Creator Mode: Scriptwriting, viral hooks, storytelling frameworks, and marketing copy.
• Reading Mode: Deep-dive document analysis, thematic breakdowns, and chapter synthesis.
• General Mode: Fast, unrestricted, all-around conversational intelligence.

2. 📄 Interactive Split-Screen Document Co-Pilot

• Side-by-Side Analysis: View the uploaded document (PDF/DOCX/TXT) on the left while conversing with Sona AI on the right.
• Direct Paragraph Citations: Click any citation badge in the AI response to auto-scroll and highlight the exact source paragraph in the document pane, eliminating LLM hallucinations.

3. 🎙️ Real-Time Conversational Voice Call

• Low-latency, hands-free conversational voice mode powered by web speech synthesis and Groq LPU processing.
• Live audio waveform visualizer, instant transcription, and automatic conversation turn-taking.

4. 🎧 Multimodal Speech-to-Text Ingestion

• Upload audio/video lectures, meetings, and voice memos.
• Transcribed with high accuracy via AssemblyAI and automatically indexed into the active workspace context.

5. 🛡️ Enterprise Admin Command Center & Maintenance Mode

• Live Metrics: Real-time analytics for users, active sessions, audio logs, and API health.
• User Management: Search, view, and promote user permissions with server-side role gating.
• Audit Logging: Comprehensive tracking of administrative actions and security events.
• Maintenance Mode with Admin Bypass: Put the application into maintenance mode for public users while administrators retain full access for testing and verification.

---

🏗️ Technical Architecture

graph TD
    User([Web Client / User]) -->|HTTPS / Next.js 14 App Router| NextServer[Next.js Application Server]
    
    subgraph Frontend Architecture
        NextServer --> AppUI[Chat Workspace & Split-Screen Co-Pilot]
        NextServer --> VoiceUI[Real-time Voice Call Interface]
        NextServer --> AdminUI[Admin Command Center]
    end

    subgraph Security & Middleware
        NextServer --> Middleware[middleware.ts: Auth Gate + Maintenance Mode]
        Middleware --> AdminBypass{Is Admin?}
    end

    subgraph AI & Processing Pipeline
        NextServer -->|Ultra-Low Latency Inference| Groq[Groq LPU Cloud: llama-3.3-70b-versatile]
        NextServer -->|Audio / Video STT| Assembly[AssemblyAI Speech-to-Text Engine]
        NextServer -->|Document Parser| DocEngine[PDF / DOCX / TXT Extraction & Chunking]
    end

    subgraph Persistence & Security Layer
        NextServer -->|RLS Protected Queries| SupabaseDB[(Supabase PostgreSQL)]
        NextServer -->|Secure Storage Buckets| SupabaseStorage[Supabase Private Storage]
    end

Loading

---

💻 Tech Stack

Layer	Technology	Purpose
Frontend Framework	Next.js 14 (App Router)	Server Components, Streaming SSR, Optimized routing
Language	TypeScript	Strict type-safety across all components, APIs, and schemas
Styling	Tailwind CSS + Lucide Icons	Responsive modern design, smooth transitions, glassmorphism
Database & Auth	Supabase (PostgreSQL)	Row Level Security (RLS), OAuth/Email Auth, Realtime DB
AI Inference	Groq Cloud LPU™	Sub-second inference via llama-3.3-70b-versatile
Speech Processing	AssemblyAI + Web Speech API	High-accuracy transcription & bidirectional voice loop
Document Processing	pdf-parse, mammoth	Robust parsing of PDFs, Word docs, and plain text

---

⚡ Quick Start & Local Setup

1. Prerequisites

• Node.js v18.17.0+ (or v20+)
• npm or pnpm
• A Supabase project
• A Groq Cloud API Key
• An AssemblyAI API Key (optional, for audio transcription)

2. Clone the Repository

git clone https://github.com/orbit-flow-labs/SUNO-AI.git
cd SUNO-AI

3. Install Dependencies

npm install

4. Configure Environment Variables

Copy .env.example to .env.local:

cp .env.example .env.local

Fill in your configuration:

# Supabase Configuration
NEXT_PUBLIC_SUPABASE_URL=https://your-project.supabase.co
NEXT_PUBLIC_SUPABASE_ANON_KEY=your-anon-key
SUPABASE_SERVICE_ROLE_KEY=your-service-role-key

# AI Inference (Groq Cloud LPU)
GROQ_API_KEY=gsk_your_groq_api_key

# Speech-to-Text (AssemblyAI)
ASSEMBLYAI_API_KEY=your_assemblyai_key

# Public App URL
NEXT_PUBLIC_APP_URL=http://localhost:3000

5. Run Database Migrations

Execute the migrations in your Supabase SQL Editor in order:

1. supabase/migrations/0001_init.sql (Tables, RLS policies, Storage buckets, Core triggers)
2. supabase/migrations/0002_admin_enhancements.sql (Admin audit logs, composite indexes, avatar triggers)

6. Start the Development Server

npm run dev

Open http://localhost:3000 in your browser.

---

👑 Granting Administrator Privileges

To grant an account admin access (for accessing /admin and bypassing maintenance mode):

UPDATE profiles 
SET is_admin = true 
WHERE email = 'admin@yourdomain.com';

---

📚 Detailed Documentation

For judges, developers, and evaluators, detailed sub-documentation is available:

• 📖 Hackathon Judges Guide — Comprehensive review guide, evaluation criteria, and feature demonstrations.
• 🏛️ System Architecture — Deep dive into the data flow, AI fallback strategies, and security design.
• 🚀 Deployment Guide — Step-by-step instructions for production deployment on Vercel and Supabase.
• 🔒 Security & Performance — Analysis of RLS enforcement, zero secret leakage, and sub-second latency optimizations.

---

🛡️ Security & Privacy Assurance

• Zero Secret Leakage: No private API keys or service role tokens are ever transmitted to the client bundle or printed to console logs.
• Row Level Security (RLS): Every database table is protected with strict PostgreSQL RLS policies ensuring users can only read and modify their own data.
• Server-Side Authorization: All administrative endpoints (/api/admin/*) independently verify user session and is_admin status on the server before processing.

---

📄 License

This project was built for hackathon demonstration. All rights reserved by Christian01-del and reactjay.