Feedy EXT - Brand Sentiment Analysis

Feedy EXT - Brand Sentiment Analysis

Inspiration

Social media management is a time-intensive task for brands, with community managers struggling to keep up with the overwhelming volume of comments across multiple platforms. Many users are most active during nights and weekends—times when social media teams are typically unavailable. This gap in responsiveness leads to missed engagement opportunities, unaddressed negative feedback, and inefficiencies that can harm a brand’s image. Feedy EXT was born from the need to provide brands with a scalable, automated, and intelligent solution to ensure seamless social media management and engagement, even outside office hours.

What It Does

Feedy EXT is a Chrome extension designed for social media and PR teams, offering a unified platform to manage comments from LinkedIn, Google My Business, TikTok, Twitter, Facebook, and Instagram. It centralizes brand-related user feedback into an actionable dashboard that:

Scalable and modular, with full support for multiple brands (including CRUD functionality for managing companies);
Analyzes comments, classifying them as positive, negative, or neutral (using the Prompt API on-device, with a fallback to Gemini Pro 1.5 via Vertex AI for processing);
Actions: automates liking of positive/neutral comments;
Actions: suggests tailored responses for negative feedback (using the context from the original post, company context, and user reply—executed at scale with Gemini 1.5 on the cloud for one or many comments);
Insights at brand level: Ensures privacy and scalability by ingesting all analyzed data in real time into Google BigQuery using Cloud Run. This architecture enables brands to securely consolidate insights from multiple social platforms and monitor key metrics. At the brand level, teams can access a comprehensive dashboard that reveals sentiment trends, engagement data, and actionable insights—all in one place;

With 24/7 monitoring and AI-driven insights, Feedy EXT helps brands stay connected with their audience, enabling better engagement and informed decision-making.

How We Built It

Sentiment Analysis: Feedy EXT leverages Chrome’s Prompt API to perform on-device sentiment analysis of social media comments;
Cloud-Based Processing: In cases where the API isn’t compatible, a fallback system uses Vertex AI for cloud-based processing via Gemini Pro 1.5;
Data Storage & Visualization: Data is securely stored in Google BigQuery and visualized through Looker Studio dashboards;
Modular Architecture: The extension’s modular and scalable architecture ensures flexibility for multi-brand support and seamless integration into diverse workflows;

By combining Chrome’s built-in capabilities with advanced cloud processing, Feedy EXT delivers both speed and comprehensive insights.

Challenges

Developing Feedy EXT presented several challenges, including:

Integrating multiple social media platforms into a unified tool;
Balancing on-device processing with cloud-based fallbacks to optimize performance and privacy;
Ensuring a user-friendly interface while managing the complexity of AI-driven features;
Maintaining data security and scalability to handle diverse brand requirements;

Accomplishments

The team is proud to have created an adaptable and scalable solution that bridges the gap between AI-powered analysis and intuitive user interfaces. Key milestones include:

Seamlessly combining on-device and cloud-based processing for robust performance;
Developing a lightweight Chrome extension that simplifies complex social media tasks;
Delivering actionable insights through an intuitive dashboard that supports multi-brand management;

What We Learned

The development of Feedy EXT highlighted the importance of balancing advanced AI capabilities with simplicity in user experience. The team gained valuable insights into integrating Chrome’s Prompt API, maintaining data security, and optimizing cross-platform compatibility. Feedback from early testers underscored the importance of a seamless and magical user experience to drive adoption. While the promise of on-device LLMs offers exciting opportunities—such as low latencies, near-zero costs, and full privacy since data doesn't travel over the internet—we're not quite there yet. On the other hand, products that achieve success pay attention not only to their AI engines but also to UI/UX. The simpler and more magical the user experience feels, the better the adoption and satisfaction.

What’s Next (Roadmap)

The future of Feedy EXT focuses on enhancing its capabilities:

Full Automation with a Human Touch: Introducing features that automate checking brands posts on social networks + routine tasks while keeping critical decision-making in human hands;
Expanded Platform Integration: Extending support to additional social platforms for broader coverage; (Google My Business, Facebook, Instagram, TikTok, X/Twitter);
Natural Language Queries: Adding conversational capabilities to the dashboard (Query your data in natural language with Gemini assistance), allowing users to ask questions and receive insights in real time;

Team

🙌 Big thanks to Google and Devpost for this incredible opportunity! A huge shoutout to the amazing team for their fantastic work—it's been such a pleasure collaborating on this project! 👏🎉