How Baidu Reimagined Its App Personal Center with AI: Design Strategies and Results

Preface

Rapid AI development creates new possibilities for product design. Baidu, as an AI industry leader, continuously explores deep integration of AI into products to improve user experience. This article uses Baidu App's personal‑center AI reconstruction project as a case study, showing how AI recommendation and dialogue capabilities can transform a traditional static personal center into an efficient, intelligent AI‑powered hub.

Old Personal Center

The legacy personal center faced three core issues: (1) many functional entry points with low screen efficiency, (2) deep operation layers for some functions, and (3) unintuitive asset update reminders. The goal was to streamline the structure, surface high‑frequency functions, and strengthen information prompts, using AI to address these challenges.

AI Redesign of Personal Center

No existing product has applied AI to reconstruct a personal center, so designers needed to innovate from 0‑1. The redesign focuses on five aspects—framework, structure, content, operation, and dialogue—building functions, enriching details, and expanding depth to create a new AI‑enabled personal center.

1. Container Innovation Design: Building an AI‑Aware Asset Framework

We rebuilt the personal‑center framework, retaining high‑frequency function entrances based on user behavior data while removing numerous content‑recommendation cards to simplify the page. An AI‑driven recommendation container was added to host recommended content, making it easier for users to revisit.

Three page‑framework prototypes—embedded, layered, and card‑based—were evaluated. The embedded framework, which offers stronger AI integration and immersive experience, was chosen. Its header displays minimal information, the middle retains frequent functions, and the bottom area delivers AI‑recommended content.

2. Proactive Intelligent Recommendation Based on User Behavior

Within the recommendation container, the personal center leverages Baidu App usage data to intelligently suggest content and functions. When a user's asset changes—e.g., a downloaded series receives an update—the center proactively prompts the user to continue watching.

For potential user needs, such as system cache cleanup after prolonged use, the center proactively recommends the relevant operation.

3. Building a Card‑Based Content Structure for All Services

We extracted key information from each service and aggregated it into cards, ensuring flexibility for AI recommendation. Cards use conversational prompts to engage users, offering actions such as jumping to a resource, updating status after return, or performing one‑click operations like cache cleaning.

Through multiple iterations, three card styles were explored; the final 3.0 version adopts a unified template that simplifies layout, reduces interaction steps, and groups functions for clearer understanding.

Over 20 services were covered with 10 types of aggregated cards, defining three card‑operation/update patterns. More services are being integrated.

4. Content “Breaking‑Out” Prompts to Boost Operational Efficiency

AI‑driven recommendation cards dramatically reduce user steps. For example, accessing a downloaded video drops from four steps to one click via a card, which also updates the playback progress upon return.

Functional operations like cache cleaning are similarly streamlined: a single “Clear” button on a card completes the task and provides real‑time feedback, while other settings (dark mode, font size) are also made one‑click accessible.

5. Expanding Dialogue Scenarios to Address Long‑Tail Needs

Beyond recommendation cards, users can initiate queries via a dialogue interface. Clicking “Ask” or swiping up opens an AI assistant that first presents quick commands, then surfaces relevant function cards (e.g., “Clear cache” when the device is sluggish). After completing the action, the dialogue returns to the user with updated status.

The dialogue flow defines four interaction/update patterns: “Jump and return update”, “Jump to view”, “Current operation”, and “Current operation with continued dialogue”. Future iterations will expand scenario coverage.

AI Personal Center Experiment Results

After launching the AI version, key metrics for Baidu App and the personal center improved significantly, delivering positive returns. The full rollout occurred at the end of June, inviting user feedback.

Reflection and Outlook

AI‑enabled products are a market trend. Designers must first clarify AI capabilities and advantages, then boldly innovate beyond existing references, redefining frameworks, structures, and content. Regardless of AI use, core design thinking remains user‑centric; AI serves as a powerful tool to generate new perspectives and enhance experiences. Ongoing efforts will focus on deepening AI perception, delivering more precise personalization, and building a comprehensive AI assistant across Baidu App.