TabSense

Introduction to Tabsense
key features
Thematic clusters of tabs with one click bookmark option
Ask AI questions about your tabs
Hybrid inference choice according to user preference
Customization options to finetune user experience
All your bookmark at your fingertips
Detect and close duplicate tabs
Group clustered tab with just a click

💡 Inspiration

Imagine trying to prepare for an important exam or conduct deep research — you open dozens of tabs, each with promising resources, videos, and articles. Soon, your browser becomes an endless sea of clutter, and instead of learning or working, you’re hunting for the right tab. For students, researchers, and knowledge workers, this isn’t just an inconvenience — it’s a daily productivity drain.

Enter TabSense — your intelligent tab companion. Inspired by the frustration of spending hours comparing YouTube channels and websites while preparing for , TabSense was born to turn chaos into clarity. It uses AI to summarize, cluster, and organize your tabs into meaningful themes, helping you focus on learning, not searching. Just like a personal research assistant, it understands your browsing context, saves your time, and lets you query your open knowledge base naturally — so you can think in ideas, not URLs.

⚙️ What it does

TabSense redefines how knowledge workers and researchers interact with information on the web. Instead of treating tabs as static pages, it transforms them into a living, searchable knowledge base powered by AI.

At its core, TabSense leverages Firebase AI Logic for hybrid inference — enabling users to seamlessly summarize, cluster, and query their tabs. Here’s what makes it powerful:

Thematic Clustering: Tabs are automatically grouped into coherent themes (e.g., “Machine Learning Resources” or “DeFi Articles”), allowing users to instantly see the bigger picture.
Bookmarks and Tab Grouping: Save themed clusters as bookmarks, creating a knowledge base that can be queried later. Users can also group and reopen bookmarked tabs using Chrome Extension APIs.
Natural Language Querying: Ask questions like “Which tabs discuss Ethereum Layer 2?” and receive synthesized answers directly from your open tabs.
Adaptive AI Modes: Choose between local model execution for complete privacy or cloud-based models for enhanced performance.
Customizable Models: Select preferred AI models and fine-tune parameters such as temperature, context length, and precision level for personalized results.

Built on Firebase AI Logic, TabSense offers a thoughtful privacy-performance tradeoff — process data locally within the browser for maximum privacy, or leverage cloud inference for faster, more accurate responses.

In essence, TabSense is an intelligent browser companion that understands your research flow — helping you focus on insights, not on managing tabs.

🧩 How we built it

The innovation behind TabSense lies in its hybrid AI-driven architecture. By leveraging Firebase AI Logic and the Prompt API, we’ve created an intelligent system that:

Analyzes and chunks tab data for optimal processing
Applies thematic clustering through prompt-based design
Utilizes Chrome Extension APIs to group tabs, save themed bookmarks, and manage them effortlessly
Implements hybrid inference via Firebase AI Logic for natural language querying and model parameter fine-tuning
Works seamlessly across any website and browsing workflow

Our development centered around two core principles: prompt-driven clustering and user personalization. Every component — from our clustering algorithms to our natural language query system — was built to empower users with a smarter, more adaptive browsing experience.

🚧 Challenges and Breakthroughs

Building TabSense meant navigating several complex technical challenges. We needed to:

Design highly effective prompts for thematic clustering and contextual tab understanding
Filter each tab’s data to retain only relevant content chunks — optimizing for both precision and token efficiency
Extract meaningful context from raw DOM data to provide accurate input for AI clustering and natural language querying
Engineer efficient chunking strategies to overcome token limitations during processing
Mitigate AI hallucinations that occasionally caused inaccurate tab groupings
Enable LLM querying of tab data with rich contextual grounding for reliable results

Through continuous experimentation, we successfully integrated Firebase AI Logic into the workflow, establishing a hybrid inference system. Iterative refinement of prompt designs drastically reduced AI hallucinations and led to more coherent thematic clustering.

This journey also marked our first hands-on experience with Chrome Extension APIs, which proved pivotal in realizing the full functionality of TabSense.

🏆 Accomplishments that we're proud of

Built a hybrid AI pipeline combining local and cloud inference through Firebase AI Logic
Designed and fine-tuned prompts that significantly minimized AI hallucinations
Developed an intelligent tab clustering and bookmarking system powered by contextual inference
Achieved seamless Chrome Extension integration for grouping, saving, and retrieving tabs
Created a query-based knowledge system that transforms browser tabs into an active, searchable workspace

Each of these milestones reinforced our mission for TabSense — helping users shift their focus from tab management to insight discovery.

📚 What we learned

Building TabSense was a deep dive into the intersection of AI systems design and browser automation. Along the way, we learned:

The art of prompt engineering — how subtle structural changes can greatly improve clustering and response accuracy
Techniques for extracting meaningful context from DOM data for model input
The impact of efficient data chunking on reducing token consumption and improving inference quality
How to harness Firebase AI Logic for hybrid inference, balancing privacy and performance
The fundamentals of building Chrome Extensions and working with APIs for tab management
That iterative experimentation is essential to minimizing hallucinations and achieving consistent results

Ultimately, we discovered that creating a seamless AI-powered browsing experience requires not only technical depth but also empathy for user workflows and research behaviors.

🚀 What's next for TabSense

TabSense is evolving into an even smarter, more adaptive research companion. Our roadmap includes:

Enhancing natural language querying with deeper contextual understanding
Further reducing AI hallucinations through adaptive prompt tuning and real-time feedback
Expanding Chrome API utilization for richer tab interactions and multi-context browsing
Introducing cross-device sync and collaborative research modes
Exploring integration with broader AI ecosystems for unified knowledge discovery

By refining both its intelligence and usability, TabSense aims to redefine how researchers and professionals organize, explore, and act on information — one tab at a time.