Ai Tab Manager

AI-powered Chrome tab manager that understands intent, auto-organizes your workspace, and lets you control tabs with natural language commands.

Comment

Inspiration

I kept drowning in open browser tabs. So I built AI Tab Manager as a tool that understands tab meaning, helps surface what matters, and lets me manage tabs with natural language.

What it does

TabMind adds an intelligent Chrome side panel with:

  • Semantic search Find tabs by meaning
  • Natural language assistant Control tabs with plain
  • Context-aware conversations Assistant remembers your previous actions ("close those tabs" refers to tabs from last command)
  • Smart tab scoring Identifies "unnecessary" tabs based on age, activity, visit count, and domain priority
  • Pin management Save important tabs and auto-restore them when you open Google
  • Bot token system Power users can execute precise commands

How we built it

Chrome Manifest V3 extension with a side panel UI, React + TypeScript frontend, Dexie (IndexedDB) for local tab storage, pure TypeScript embeddings without WASM, dual AI support (OpenAI gpt-4o-mini and Claude Sonnet 4.5) with fallback, and hybrid command parsing (offline regex + optional AI).

Local-first embeddings, Cosine similarity and text vectorization in pure TypeScript (no heavy deps) Conversation context system, Persistent context with auto-summarization to reduce tokens Hybrid command parsing, Natural language to AI to tokens to Chrome APIs with local fallback Pin restore flow, Detects Google pages and restores pinned tabs in-window

Challenges we ran into

Chrome permission constraints for script execution. Robust multilingual intent parsing. Converting natural language into deterministic bot commands. Balancing semantic ranking quality with lexical fallback speed. External model-ID compatibility issues across AI providers.

Accomplishments that we're proud of

We built a fully working Chrome side panel extension that turns messy tab overload into a manageable workflow with semantic search, AI grouping, and fast tab actions. We are proud that TabMind combines local indexing and embeddings with practical automation so users can find tabs by meaning, not just exact words, while still keeping control through clear commands and one click actions. We also implemented a unified assistant that supports natural language and structured commands, plus smart pin restore and explainable unnecessary tab detection, making the product useful in real day to day browsing sessions rather than only as a demo.

What we learned

How to build reliable AI UX from free-form language to precise actions. How to combine vector search with rule-based safety and control. How to structure a Chrome extension with clear UI/background/storage boundaries.

What's next for Ai Tab Manager

Multi-device sync via Chrome Sync, automatic tab archiving to the cloud with on-demand restore, semantic search across browsing history, visual tab clustering (t-SNE/UMAP), cross-browser support (Chrome, Firefox, Edge, Safari), shared team workspaces, scheduled tab cleanup automation, and voice control via Web Speech API.

Built With

Share this project:

Updates