Analytics · Artificial intelligence
ChatMood
The audience pulse, live. Sentiment analysis turning Twitch conversations into useful information.
The whole
product.
ChatMood solves a concrete problem for streamers and brands: understanding what the audience is feeling and discussing during a live broadcast without manually reading thousands of messages. The platform connects to the user's Twitch channel via OAuth, captures chat messages in real-time using tmi.js, and sends message batches to OpenAI's GPT-4o-mini model for structured sentiment analysis, topic extraction, FAQ detection, emote analysis, and actionable insights. The system includes a live dashboard with interactive visualizations — sentiment gauge, temporal word cloud, engagement charts via Recharts, AI insights carousel, audience questions panel, and emote analysis. Beyond real-time monitoring, it offers configurable alerts (by sentiment, engagement, or topic) with email and webhook notifications, data export, and analysis history. Subscription billing is handled by Stripe. Built with Next.js 14, TypeScript strict, MongoDB, and a glassmorphic dark-theme UI.
Solution scope
AI sentiment analysis, live dashboards, channel comparison and Stripe subscriptions.
How it works
in practice.
Capabilities connecting the user experience to operational needs.
Real-Time AI Chat Analysis
Pipeline connecting tmi.js to GPT-4o-mini: messages are batched, filtered, and analyzed for sentiment, topics, questions, emotes, and actionable insights.
Live Dashboard
13 interactive widgets: sentiment gauge, temporal word cloud, engagement charts, AI insights carousel, audience questions, and emote analysis with 5-second polling.
Stripe Subscription Billing
Subscription checkout, billing portal, webhooks, and plan enforcement through Stripe.
Configurable Alert System
Rules based on sentiment, engagement, or topics with email and webhook notifications. HTTPS validation, private IP blocking, 5-min cooldown.
Multi-Provider Authentication
NextAuth with credentials + Twitch OAuth, email verification, password reset, JWT in httpOnly cookies.
Internationalization (EN/PT)
Custom i18n system via React Context with 800+ translation keys covering the entire interface.
Behind
the interface.
Next.js App Router (Feature-based / Collocated)
Full-stack architecture based on Next.js 14 App Router where API routes and pages coexist in a feature-oriented structure. The backend uses API Routes as lightweight microservices, with a lib layer for business logic and Mongoose models for persistence. Custom server.js manages WebSocket connections and orphan session cleanup.
Solution components
- Presentation (Pages, Components, Hooks, Context)
- API Routes (RESTful Controllers)
- Business Logic (rate-limit, plan-limits, alert-checker)
- Persistence (Mongoose, MongoDB)
- External Integrations (Twitch, OpenAI, Stripe)
Interface & application
Libraries
Data
Infrastructure
Integrations
Explore
the details.
Actual project screens. Select an image to enlarge it and browse the gallery.
