Making the Browser Feel Personal

Explored how customization could make Neo feel personal and familiar across different working styles — through vertical tabs, split tabs, accent colors, and a personalized AI-powered feed.

Project

Browsers become personal through use. Over time, people develop their own systems for working, reading, saving, comparing, and coming back to information. Some users keep dozens of tabs open as a working memory system. Some prefer a clean horizontal tab bar. Others rely on sidebars, split views, visual cues, or daily rituals around what they read.

As we designed Neo, personalization became an important part of making the browser feel useful and familiar. The challenge was not simply adding more settings. It was designing a browser that could adapt to different working styles while still feeling like one coherent product.

This work focused on four areas of customization: vertical tabs, split tabs, accent colors, and a personalized AI-powered feed.

Goals

Support different tab behaviors without creating separate interaction models.
Make AI context predictable when multiple pages are visible.
Allow visual personalization without weakening accessibility or system consistency.
Expand personalization beyond appearance into what users read and return to every day.

Final Design

Easier tab management for users with many open tabs

Vertical tabs give users more room to scan and manage tabs, especially when they have many pages open at once. The design kept the same core tab behaviors across horizontal and vertical layouts, so users could switch between layouts without relearning how tabs work.

Split tabs with explicit AI context controls

Split tabs allow users to view two pages side by side for comparison, reference, writing, and research. Because multiple pages can be visible at once, I designed explicit context controls so users could see which page the AI was using, switch context, or include both pages when needed.

Lightweight accent color customization

Accent colors let users personalize the browser without requiring a full theme builder. I defined where accent colors could appear in the design system and tuned them across light and dark themes, so the browser could feel personal while staying readable, accessible, and visually consistent.

A personalized AI-powered feed

The custom feed helps users shape what they read every day. Instead of manually configuring every source or topic, users can describe their interests in natural language, then review and refine an AI-generated feed over time.

Design Details

One tab system, two layouts

Problem: Supporting both horizontal and vertical tabs risked creating two separate interaction models, where users would have to relearn basic browser behaviors depending on the layout they chose.
Solution: I treated horizontal and vertical tabs as two expressions of the same underlying tab system, keeping actions like opening, closing, pinning, grouping, reordering, and switching tabs consistent across both layouts.

Making AI context visible and editable

Problem: AI responses are only useful when the right context is included, but users often do not know what the AI can see or use. In a browser, this creates trust issues around pages, selections, tabs, and chat history.
Solution: I explored context controls that made attached sources visible in the side chat. Users could add page or selection context, understand what was being referenced, and include or exclude context from the conversation.

Making split context understandable

Problem: Split tabs made AI context more ambiguous because multiple pages could be visible at once, making it unclear which page the AI should use when responding.
Solution: I made active context visible and adjustable, so users could see which page the AI was using, switch context when needed, or include both pages for comparison.

Making AI workflows easier to discover and reuse

Problem: Many AI workflows depend on repeated instructions: summarize this, rewrite in this tone, extract key points, compare options. Rewriting those prompts every time slowed users down, and users often struggled to discover effective prompts in the first place.
Solution: I designed a Snippets Gallery where users could browse, save, and reuse curated prompts for common tasks. By making proven workflows discoverable and easy to personalize, the gallery helped users get started faster, improve consistency, and unlock more value from AI across everyday browsing tasks.

Personalization without breaking the system

Problem: A personalized feed could become too much work if users had to manually configure every topic, source, keyword, and preference through settings.
Solution: I designed a conversational flow where users could describe what they wanted to follow, then review and refine the AI-generated feed over time.

Next Steps

Tabs as artifacts of intent
The next step is to move beyond thinking of tabs as the primary unit of interaction. Whether a user chooses horizontal tabs, vertical tabs, or split view, tabs are ultimately artifacts of what they are trying to do. The larger opportunity is helping the browser understand the task behind the tabs: what the user is comparing, researching, writing, buying, planning, or trying to remember. A more intelligent browser should not only organize pages. It should help infer the user's intent and surface the right assistance at the right moment.
Agents in action-oriented workflows
For workflows that require action — booking, purchasing, creating documents, filling forms, or completing multi-step tasks — the browser will likely need agent capabilities built directly into the experience. That introduces a new layer of design challenges: how much control the user keeps, how the agent explains what it is doing, when it asks for confirmation, and how the browser makes automated actions feel trustworthy. Customization was one step toward making Neo feel more personal. The larger opportunity is making the browser more adaptive, contextual, and action-oriented.

Key Takeaways

Balance customization with consistency
Personalization is not just about adding more options. It is about helping the product adapt to different working styles while preserving a coherent system. The main design challenge was deciding where flexibility should end and consistency should begin. Vertical tabs, split tabs, accent colors, and custom feeds all gave users more control, but each feature needed clear boundaries to keep the experience understandable.
Prioritize requests through product direction
Personalization was one of the most common requests from early users, which made it an important area to explore. But the broader product question was not just “What settings do users want?” It was “What should a browser become in the age of AI?”
This work helped balance immediate user feedback with longer-term product direction. Some customization features made the browser feel more familiar and adaptable. Others opened up bigger questions about intent, context, and how AI should participate in everyday browsing workflows.