OmniRepute

Leverage AI to gain deep insights about your brand's performance across multiple data sources.

Comment

Inspiration

In today’s hyperconnected world, brands live and die by public perception. A single viral post or negative headline can reshape reputation overnight. Yet, tracking and understanding brand sentiment across countless platforms remains an enormous challenge.

We built OmniRepute to bridge that gap — an AI-powered platform that helps organizations truly understand how their brand is perceived. Our goal was to create a tool that doesn’t just monitor mentions, but interprets them — providing actionable insights that drive better communication, trust, and strategy.

What it does

OmniRepute transforms raw social and news data into clear, data-driven reputation intelligence. By aggregating information from Reddit and GDELT news sources, it evaluates how audiences perceive a brand and provides:

  • Reputation Scoring (0–100) with detailed rationale
  • Key Insights summarizing brand sentiment
  • AI-Generated Strategies for improving public perception
  • Sentiment Breakdown — what people love and what they criticize
  • Complaint Response Templates — AI-suggested replies for common issues

The result is a unified dashboard that lets brands see themselves through the eyes of their audience — clearly, accurately, and in real time.

How we built it

We combined a modern, cloud-native architecture with AI-driven analysis to ensure scalability, performance, and actionable insights.

Frontend

The frontend was built using React 18, TypeScript, and Vite to deliver a responsive and fast user experience. Tailwind CSS ensures accessibility and visual clarity, while components such as BrandInputForm, AnalysisDisplay, and ScoreCard guide users through each analysis.

Backend

The backend uses Node.js and Express as the central API layer connecting the frontend, AI, and database. It features:

  • CORS-enabled endpoints for secure communication
  • Structured AI response schemas
  • Integration with Google Gemini (Vertex AI) for analysis generation
  • BigQuery for efficient data querying and management

Data Pipeline

Data ingestion and synchronization are handled through Fivetran connectors:

  • Reddit Connector for fetching posts, comments, and sentiment trends
  • GDELT Connector for processing global news and media coverage

These connectors support incremental updates, rate limiting, and configurable lookback periods, ensuring timely and reliable data collection.

AI Integration

We integrated Google Gemini AI to interpret brand mentions, extract key insights, and generate strategic recommendations. Gemini powers:

  • Reputation scoring and rationale generation
  • Identification of “what users love” and “what users hate”
  • Creation of actionable improvement strategies and complaint responses

Deployment

The entire platform runs in Docker containers deployed on Google Cloud Platform, with BigQuery, Vertex AI, and Cloud Storage forming the data backbone for analysis and retrieval.

Challenges we ran into

  • Data Variability: Social and news data differ greatly in tone, structure, and context. Normalizing them while preserving meaning was complex.
  • AI Interpretation Accuracy: Achieving consistent sentiment analysis across diverse industries required careful prompt engineering and iteration.
  • API Rate Limits: Handling Reddit and GDELT rate limits while ensuring freshness of data demanded optimized connector design.
  • Explainability: Making AI-generated insights interpretable and actionable was key to user trust.

Accomplishments that we're proud of

  • Built and deployed a full-stack, AI-driven reputation analytics platform integrating multiple data sources.
  • Implemented real-time sentiment tracking with transparent reputation scoring.
  • Designed an intuitive, elegant interface that visualizes complex brand analytics effortlessly.
  • Developed custom Fivetran connectors for Reddit and GDELT, enabling automated and continuous data ingestion.
  • Demonstrated how AI and data engineering can turn unstructured content into meaningful brand intelligence.

What we learned

  • Context is everything. AI insights are only as strong as the context we provide. Tailoring Gemini prompts for brand analysis was essential.
  • Clean data drives clarity. Investing in data normalization and validation significantly improved insight accuracy.
  • Design matters. Presenting analytics in an intuitive, visually clean interface enhanced usability for non-technical users.
  • Connectors accelerate innovation. Leveraging Fivetran and Google Cloud streamlined data flow and deployment, reducing infrastructure overhead.

What's next for OmniRepute

Looking ahead, we plan to:

  • Introduce multi-brand comparison dashboards for competitive benchmarking
  • Expand data sources to include X (Twitter), YouTube, and Google Reviews
  • Implement trend forecasting for early detection of reputation shifts
  • Add custom alert systems for sentiment anomalies and crisis detection
  • Launch a beta program with marketing and PR teams for real-world testing and feedback

Our mission remains to empower brands with intelligent, real-time reputation insights that transform how they engage with the world.

Share this project:

Updates