• 

• 

• 

• 

• 

Inspiration

Customer relationships are the heart of every business — but managing customer sentiment and support manually takes time, effort, and empathy. AURA was inspired by the need to make support more human and more intelligent. I wanted to build an AI-powered assistant that could analyze customer emotions, automate responses, and give managers insights on what customers truly feel — all in real time.

What it does

AURA – AI Customer Relationship is a smart customer support dashboard that:

Analyzes messages for sentiment (positive, neutral, or negative)

Detects urgency automatically (“refund asap”, “angry”, etc.)

Auto-generates intelligent bot responses using context and FAQs

Displays a manager dashboard with analytics — including top issues, customer churn risk, and sentiment trends

Allows agents to add FAQs dynamically to train the assistant

Includes API key–based access control for admin endpoints

How I built it

Frontend: Next.js (React) + Tailwind CSS for a fast, modern interface

Backend: FastAPI with SQLModel for RESTful APIs

Database: SQLite (locally) / PostgreSQL (for production on Render)

AI / NLP: VADER Sentiment for emotional tone analysis

Hosting:

Frontend → Vercel

Backend → Render

Integrated the two via environment variables (NEXT_PUBLIC_API_BASE)

Challenges I ran into

Debugging CORS and deployment issues between Render and Vercel.

Keeping sentiment analysis lightweight but accurate.

Managing in-memory state during local development before adding persistence.

ESLint + TypeScript configuration errors during build on Vercel.

Designing clear data flow between chat, analytics, and FAQ modules.

Accomplishments that I’m proud of

Fully functional AI-driven CRM that runs on live servers (Vercel + Render).

Real-time sentiment and urgency detection pipeline.

Integrated analytics dashboard that visualizes live customer mood data.

Clean, minimalist UI using Tailwind CSS.

Seamless multi-environment deployment with proper API key security.

What I learned

How to build scalable full-stack systems combining FastAPI + Next.js.

How to integrate AI/ML features directly into customer workflows.

The importance of clean REST design and proper CORS configuration.

How to balance UX simplicity with backend intelligence.

What’s next for AURA – AI Customer Relationship

Integrate LLM-based conversational intelligence for more natural responses.

Add email / Slack connectors for real-world customer support pipelines.

Use Neon/PostgreSQL for persistent data across sessions.

Train sentiment models on domain-specific datasets for higher accuracy.

Build a multi-tenant dashboard for teams and enterprise clients.

Built With

• fastapi
• next.js
• postgresql
• pydantic
• python
• react
• render
• sqlalchemy
• sqlmodel
• tailwind-css
• typescript
• vader-sentiment
• vercel