Overview

Bookar is an AI-powered knowledge platform built around a simple idea: bookmarks should remain useful long after they are saved. Rather than acting as a traditional bookmark manager, it transforms saved links into searchable, AI-understandable knowledge that can be accessed from both the web and modern AI assistants.

The platform combines a web dashboard, browser extension, cloud synchronization, and a standalone MCP server to create a persistent knowledge layer that works across devices and workflows.

Problem

Most bookmark managers become long-term storage rather than active tools. As collections grow, finding useful resources becomes increasingly difficult, and valuable knowledge is often forgotten.

Modern AI assistants also have no awareness of the resources users have collected over time, forcing the same context to be rediscovered or re-explained repeatedly.

Solution

Bookar enriches every bookmark with AI-generated context instead of storing only a title and URL. Descriptions, metadata, and searchable information make bookmarks easier to rediscover while exposing them through the Model Context Protocol (MCP), allowing compatible AI assistants to access a user's curated knowledge securely.

The goal is to shift bookmarks from passive storage into an active knowledge system that supports learning, research, development, and everyday productivity.

User Experience

The experience is designed around speed and clarity. Users can save bookmarks directly from the browser extension, manage them through a responsive dashboard, and organize information without unnecessary complexity.

A clean dark interface, minimal navigation, fast search, and thoughtful visual hierarchy keep attention on content instead of interface elements. Device linking and cloud synchronization ensure the same knowledge is available wherever the user works.

Architecture & Development

Bookar is structured as a monorepo with separate applications for the web dashboard and the standalone MCP server, allowing both services to be deployed independently while sharing the same Supabase backend.

The frontend is built with Next.js and TypeScript, while Supabase provides authentication, database services, and secure user storage. NVIDIA NIM powers AI-generated bookmark descriptions and metadata enrichment. The browser extension communicates with the web platform for synchronization, while the dedicated MCP server exposes bookmark knowledge to compatible AI clients. Both applications are deployed independently on Vercel to simplify scaling and maintenance.

Challenges & Iterations

One of the primary engineering challenges was separating the user-facing application from the MCP service without duplicating backend infrastructure. Establishing a shared authentication model while keeping deployments independent improved maintainability and future scalability.

Additional iterations focused on reducing interface complexity, refining the information architecture, improving bookmark discovery, and creating an experience that feels closer to a productivity application than a conventional bookmark manager.

Results & Impact

Bookar demonstrates the design and development of a complete AI-powered web application spanning frontend engineering, backend architecture, browser extensions, authentication, cloud infrastructure, API integration, and modern AI workflows.

The project showcases product thinking alongside technical implementation by solving a familiar problem through a more intelligent and extensible approach to personal knowledge management.

What I Learned

Building Bookar strengthened my understanding of scalable application architecture, authentication systems, AI integration, browser extension development, and API design. It reinforced the importance of separating responsibilities between services while maintaining a consistent developer experience.

The project also deepened my appreciation for product thinking, maintainable codebases, deployment strategies, performance optimization, and designing software around real user workflows rather than isolated features.

Future Roadmap

Future development will focus on semantic search, intelligent collections, collaborative workspaces, advanced filtering, richer AI retrieval capabilities, broader MCP compatibility, additional AI providers, improved browser integrations, offline support, and expanded productivity workflows that help users turn saved knowledge into actionable insights.