Tailor

Tailor is an AI powered browser extension that transforms Chrome into the browser you want, installing dozens of extensions. Instead of adapting yourself to your browser, your browser adapts to you.

Comment

Inspiration

Every browser is built for the average user. But nobody browses the internet the same way.

Some people want fewer distractions. Others want smarter workflows, cleaner interfaces, or information that websites don't provide. Yet making these changes usually means searching through dozens of Chrome extensions, writing custom scripts, or switching browsers entirely.

We asked ourselves a simple question:

What if you could tell your browser what you wanted, and it could tailor itself to you?

That's how Tailor was born.

We wanted to transform the browser from a fixed interface into something personal, adaptable, and programmable by anyone. No coding experience, no hunting for extensions, and no waiting for developers to implement a feature.

Just describe what you want in plain English and watch your browser change.

What it does

Tailor is an AI-powered Chrome extension that lets anyone customize websites, create live widgets, and automate browsing tasks using natural language.

Instead of simply answering questions, Tailor's AI agent actually modifies your browsing experience.

Customize any website. Ask Tailor to hide YouTube Shorts, remove distracting elements, redesign a page, or make a website easier to read. Tailor generates the necessary JavaScript and CSS, tests the changes, and applies them directly to the page.

Create live widgets. Want live sports scores, trending news, or information from another website? Tailor can generate floating widgets that retrieve and refresh data while you browse.

Understand what's on your screen. Tailor combines visual understanding with webpage structure, allowing users to ask questions about what they're seeing and request changes based on that context, even when traditional webpage inspection isn't enough.

Automate repetitive interactions. Tailor can perform tasks such as dismissing interruptions, loading additional content, or controlling distracting page behavior.

Make customizations permanent. Changes are saved and automatically restored whenever you revisit a website. Users can manage, disable, edit, and share their customizations.

The result is a browser that gradually becomes your own.

How we built it

We developed Tailor as a Chrome extension powered by an agentic AI backend.

The extension uses WXT, React, and Chrome's Manifest V3 APIs to provide an interactive side panel and communicate with the websites users visit.

When a user requests a modification, Tailor collects relevant webpage information, including its DOM structure and context, and sends it to a FastAPI backend through WebSockets.

A Claude-powered agent interprets the request and generates JavaScript, CSS, or a custom widget. Rather than immediately executing the generated code, Tailor uses Playwright to test it in an isolated browser environment. If a test fails, the agent can inspect the error, revise its implementation, and retry.

Once validated, the extension applies the modification using Chrome's userScripts API and saves it for future visits.

We also integrated several technologies to extend Tailor's capabilities:

  • Claude: Generates, revises, and troubleshoots website modifications and widgets.
  • Gemini: Provides multimodal screen understanding by interpreting screenshots alongside webpage context.
  • Apify: Retrieves live information from websites when structured public APIs aren't available.
  • Convex: Supports real-time synchronization and a community library of browser customizations.
  • Vultr: Provides a cloud deployment target for the containerized backend.

Together, these components form a system that goes beyond generating suggestions: it turns natural-language instructions into functioning browser modifications.

Challenges we ran into

One of our biggest challenges was making AI-generated code reliable enough to execute inside a real browser.

Websites have different structures, dynamic content, constantly changing layouts, and complex JavaScript behavior. A modification that works on one website can easily fail on another.

We addressed this by designing an automated generate → test → diagnose → retry workflow using Claude and Playwright.

Another challenge was persistence. Applying a modification once is relatively straightforward, but keeping it functional across reloads and future visits requires careful management of Chrome extension permissions, background processes, and site-specific scripts.

We also had to balance flexibility with safety. Giving an AI agent the ability to write executable code introduces risks. We designed Tailor around site-scoped modifications, user-controlled toggles, and pre-execution testing. These measures improve reliability and user control, although they do not eliminate every security risk.

Finally, combining DOM inspection with screenshot-based understanding required two different representations of the same webpage. We used Gemini's visual capabilities to complement the structural information available through the browser.

Accomplishments that we're proud of

We're particularly proud of turning a simple natural-language request into an actual, persistent modification of a website.

Tailor doesn't just explain how to customize a page. It generates the modification, tests it, applies it, and remembers it.

We also designed a system where multiple AI capabilities work together: code generation, visual understanding, automated browser testing, and live information retrieval.

Another accomplishment was building Tailor around a simple principle: users should remain in control of their browser.

Every modification is designed to be visible, manageable, and reversible.

What we learned

Building Tailor taught us that generating code is only one part of creating a useful AI agent.

The harder problems involve understanding context, checking whether generated code works, managing permissions, recovering from failures, and ensuring that users can trust the result.

We also explored how Chrome extensions, browser automation, WebSockets, and multimodal AI can work together to create entirely new user experiences.

Perhaps our biggest takeaway was that AI becomes significantly more useful when it can safely act on a user's intentions instead of merely responding to them.

What's next for Tailor

We envision Tailor evolving into a personalized layer for the entire web.

Our next steps include improving automated safety checks, making generated modifications more resilient to website updates, and expanding the community library so users can discover and share customizations.

We also want Tailor to understand broader user preferences across websites, enabling someone to say:

"Make every website I visit cleaner, faster, and less distracting."

And have Tailor intelligently adapt the browsing experience to match.

Eventually, we want users to build and share entire browser experiences without writing a single line of code.

The internet was designed for everyone. Tailor makes it yours.

Built With

Share this project:

Updates

Submission history