Tabbit AI

Product Introduction

1. Definition: Tabbit AI is an AI-native, agentic web browser designed to function as a contextual work assistant. It is a desktop application (macOS and Windows) that integrates large language models (LLMs) directly into the browsing experience, enabling users to interact with, analyze, and automate tasks across web content, local files, and screenshots.
2. Core Value Proposition: Tabbit exists to transform the passive web browser into an active productivity partner. Its primary value is context-aware task automation, allowing users to offload repetitive research, data extraction, content synthesis, and organization workflows to an AI agent that operates across their digital workspace.

Main Features

1. Contextual AI Agent (Tabbit Agent): This is the core automation engine. Users can provide context by "@-mentioning" any open tab, uploaded file (PDF, image), screenshot, or highlighted text. The agent reads this context, formulates a plan, and executes multi-step tasks across the web, either immediately or on a scheduled basis. It utilizes a multi-agent architecture with specialized roles (Researcher, Operator, Writer, Analyst) loaded with pre-built Skills.
2. Skill Library & Marketplace: Tabbit provides over 2000 pre-configured, agentic "Skills"—modular automations for specific sites and tasks (e.g., YouTube, GitHub, arXiv, Gmail). Skills are executable commands (like @feed-trim to triage a feed) that encapsulate complex workflows. Users can also create and share custom Skills, turning recurring prompts into one-click actions.
3. Multi-Model Orchestration: The browser provides unified access to a vast array of cutting-edge LLMs, including GPT-5.6, Claude Sonnet 5, Gemini 3.1 Pro/Flash, DeepSeek V4, Kimi K3, Qwen 3.8, GLM 5.3, Doubao Seed, and MiniMax M3. This allows users to select the optimal model for cost, capability, or regional compliance per task, with routing through compliant servers in Singapore or the user's region.
4. Intelligent Tab & Context Management: Beyond traditional bookmarks, Tabbit allows organization based on project context. It can ingest and reason across "tab graves," PDFs, bookmarks, and HTML elements. Features like Agent Memory help maintain context across sessions, and tools auto-organize content (e.g., sorting ChatGPT threads by topic with @thread-tidy).
5. Privacy-First Data Handling: Tabbit is built with a local-first architecture. User data like highlights, chats, and history are encrypted on-device. Cloud sync (optional) uses compliant providers in Singapore. The company, Lumina Lab Pte. Ltd., is Singapore-registered and adheres to SOC 2® Type I standards, with a clear data policy of not selling, training on, or sharing user data.

Problems Solved

1. Pain Point: Information Overload and Context Switching. Professionals waste hours manually skimming articles, synthesizing reports from multiple tabs, extracting data from PDFs, and managing disparate digital content.
2. Target Audience: Knowledge Workers, Researchers, Content Creators, and Developers. Specific personas include: Academic researchers reviewing literature; Podcast hosts processing interview transcripts; Software engineers analyzing GitHub PRs; Digital marketers auditing subscriptions and feeds; Content creators repurposing long-form video into clips.
3. Use Cases:
   • Academic Research: Automatically saving papers from arXiv/PubMed with metadata (@paper-pin), extracting tables to TSV (@paper-flow), and performing Q&A across a personal library (@lib-ask).
   • Content Consumption & Creation: Generating a highlight reel from a 2-hour live stream (@stream-keeper), pulling HD videos with subtitles from Bilibili (@bili-hd), or translating subtitles in real-time (@cc-bridge).
   • Developer Workflow: Explaining a complex 47-file Pull Request in plain English (@pr-read), performing risk analysis for breaking changes (@ship-safe), and identifying untested code paths (@gap-find).
   • Personal Productivity: Getting a daily digest of all subscriptions (@inbox-zero), auditing and bulk-unsubscribing from paid services (@sub-audit, @bye-list), and auto-organizing email threads awaiting reply (@owe-track).

Unique Advantages

1. Differentiation: Unlike AI chatbot extensions or co-pilots that work within a single webpage (like ChatGPT), Tabbit is a full-stack agentic environment. It controls the browser itself, allowing it to navigate between sites, click buttons, fill forms, and orchestrate workflows across multiple tabs and applications autonomously. Compared to other AI browsers, its deep integration of a vast, user-accessible Skill library for top sites is a key differentiator.
2. Key Innovation: The "@ anything" contextual invocation system combined with a Skill-based architecture. This allows users to trigger powerful, multi-step automations by simply referencing any piece of digital content (a tab, file, screenshot) as context. The system's ability to rapidly integrate new SOTA LLMs (within 12 hours of launch) and its focus on compliant, regional data routing for each model provider are also significant technical innovations.

Frequently Asked Questions (FAQ)

1. Is Tabbit AI browser free to use? Yes, the Tabbit browser application is free forever. AI features come with a free weekly allowance on the Trial plan. Setting Tabbit as your default browser unlocks the Standard plan with 10x higher usage limits at no cost. A paid Pro tier is available for users requiring frequent access to the most advanced models and heavy Agent usage.
2. How does Tabbit AI handle my privacy and where is my data sent? Tabbit employs a privacy-first architecture. Data like history and chats are encrypted on your device. When you use an AI model, your prompt is sent only to the provider you selected (e.g., OpenAI for GPT), via compliant servers in Singapore or your legal region. Tabbit (Lumina Lab) does not log, relay, or store your conversation data. The company is SOC 2® Type I certified and does not sell or train on user data.
3. What can the Tabbit AI Agent actually do for me? The Tabbit Agent can automate complex, multi-step web tasks. For example, it can: research a topic across multiple sources and compile a report, monitor web pages for changes and notify you, extract specific data (like tables or comments) from a list of URLs, organize your saved content (papers, bookmarks, chats) into structured knowledge bases, and automate repetitive posting or data entry workflows on supported web platforms using pre-built Skills.
4. Can I import my data from Chrome or Safari into Tabbit? Yes, Tabbit supports one-click import of browsing history, bookmarks, saved passwords, and settings from major browsers including Google Chrome, Microsoft Edge, and Apple Safari during the initial setup process, ensuring a seamless transition.
5. What are Tabbit Skills and how do I use them? Skills are pre-built, reusable automation scripts for specific tasks. You use them by typing "@" followed by the Skill's name (e.g., @arxiv-flow) in the Agent chat. You can also attach context (like a PDF or tab) to the command. The Skill library includes utilities for sites like YouTube, GitHub, and arXiv, and users can build and share their own custom Skills.