Sentimentality | Devpost

Inspiration

Customer sentiment in telecom changes faster than any support team can react. One viral post can spread frustration before a company even knows there’s a problem. T-Mobile's challenge inspired us to imagine something different, a system that transforms scattered online conversations into a real-time Customer Happiness Index.

We asked ourselves:

What if T-Mobile could see customer emotion shift the moment it happens?
What if network issues, billing pain points, or moments of delight could be detected before they became trending problems?

Sentimentality was born from the desire to give telecom teams early visibility into customer feelings using real-time data, scalable engineering, and modern AI, all wrapped into a tool that could make an immediate real-world impact.

What it does

Sentimentality is a real-time customer sentiment intelligence platform designed to help companies like T-Mobile understand how customers feel the moment they express it.

It:

Captures live posts from Reddit, Hacker News, and X
Uses Gemini AI to analyze sentiment, extract core themes, and detect shifts in customer emotion
Identifies early signs of frustration related to network quality, customer support, billing, and device issues
Helps businesses detect negative spikes before they escalate
Highlights positive feedback and “moments of delight”
Gives teams actionable intelligence about customer happiness, complaints, and praise
Each post includes a ‘See Post / Reply’ button that links directly to the original Reddit, Hacker News, or X discussion, allowing T-Mobile teams to immediately view context and respond to customers where the conversation is happening.

Sentimentality turns raw public conversation into a continuous Customer Happiness Index that teams can act on in real time.

How we built it

We engineered a multi-stage real-time analytics pipeline:

Data Collection Layer

Ingests posts from Reddit, Hacker News, and X
Handles live updates using public endpoints
Normalizes content with a unified schema to make cross-platform analysis consistent

AI Sentiment Analysis (Gemini AI)

Performs sentiment classification (from -10 to 10)
Extracts keywords, emotional intensity, and topic clusters
Identifies trends such as rising frustration or emerging praise

Processing & Storage Layer

Laravel backend orchestrates ingestion, processing, and storage
Cloud-based database optimized for fast querying and scalable volume

Real-Time Dashboard

Visualizes trends, spikes, and sentiment distribution
Designed with accessibility, clarity, and intuitive navigation
Auto-updates to show live sentiment changes

Real-time Readiness

When the X API introduced up to 15-minute delays per query, we prioritized faster and openly accessible platforms like Reddit and Hacker News to maintain real-time performance, ensuring the system remained accurate, responsive, and reliable.

This architecture allows Sentimentality to scale and integrate future data sources with minimal friction.

Challenges we ran into

API Latency & Real-Time Constraints

X’s official API introduced significant delays, so we focused on high-speed platforms like Reddit and Hacker News to preserve real-time functionality while still including X as a supplementary source.

Data Normalization Across Platforms

Each platform provides content in different formats. We built a unified schema to ensure clean, consistent analysis.

High-Volume Sentiment Processing

Customer sentiment can spike quickly during outages or news events. Designing efficient ingestion and processing paths helped keep the system stable and responsive.

Real-Time UI/UX Challenges

Creating a dashboard that updates rapidly without overwhelming the user required thoughtful layout decisions, lightweight rendering, and intuitive data grouping.

LLM Optimization

We learned to optimize Gemini’s processing to manage speed, tokens, and cost, while keeping sentiment analysis accurate and insightful. Each challenge pushed us to build a more robust, scalable system.

Accomplishments that we're proud of

Built a fully functional real-time sentiment intelligence system in a short hackathon timeframe
Designed an intuitive dashboard that simplifies complex data into actionable insights
Leveraged Gemini AI to extract meaningful customer emotion at scale
Developed multi-source ingestion that adapts to API limitations
Demonstrated a solution that could be deployed and used by T-Mobile’s care, engineering, and social teams immediately
Implemented a ‘See Post / Reply’ action that deep-links to the original thread, enabling fast response workflows for T-Mobile’s care and social teams.

What we learned

Real-time systems require flexibility, APIs fail, data formats change, and platforms behave unpredictably
AI-driven sentiment analysis becomes far more powerful when paired with proper data cleaning and structure
Telecom sentiment is highly dynamic, and early detection matters
Clear visualization can make complex analytics understandable and actionable
Building a production-feel system requires combining engineering, UX design, and storytelling

What's next for Sentiment Insight

Network Issue Spike Detection

Automatically identify phrases like “no service,” “slow 5G,” “dropped calls,” and visualize them as real-time outage indicators.

Predictive Churn Modeling

Use sentiment trends + behavioral patterns to detect potential customer churn.

Topic Clustering & Root Cause Analysis

Let teams quickly understand why sentiment is changing.

Deeper T-Mobile Keyword Tracking

Coverage, billing flows, device issues, Magenta plans, customer care operations.

App Store, YouTube, TikTok & Web Review Ingestion

Expand beyond text-based communities.

Mobile Alerts for Team Leads

Push notifications for sentiment spikes or emerging issues.

Sentimentality is designed to grow into a robust customer experience intelligence platform with massive real-world impact, not just a hackathon prototype.