How Baidu App’s “Pinch‑to‑Summarize” Uses AI and Design Thinking to Boost User Efficiency

Introduction

As AI technology advances, users expect faster, more accurate, and personalized information retrieval. Designers face the challenge of leveraging design thinking and AI to help users obtain information efficiently while creating intelligent new functions.

This article shares how Baidu App’s recent “Pinch‑to‑Summarize” feature combines AI capabilities with design thinking to improve product intelligence and user experience, offering valuable references for designers focused on AI.

Background

Baidu App’s core scenarios suffer from multiple search results, inefficient long‑text reading, and slow video information extraction, leading to low user efficiency. To address this, AI is used to summarize content, creating a universal experience chain: Pinch‑pre‑guide → Pinch trigger → Content loading → Result display , aiming to improve information acquisition speed.

Design Exploration Dimensions

Highlight creation – building an innovative chain experience

Intelligent perception – crafting an AI‑driven design language

Information efficiency – structured layout to promote information acquisition

1. Building an Innovative Chain Experience

1.1 Gesture Innovation – Two‑Finger Pinch Trigger

To support all Baidu App scenarios, two trigger methods were compared: a functional entry and a gesture trigger. The functional entry adds an AI‑summary button to the universal framework, while the gesture analysis focused on the “two‑finger pinch” gesture, ensuring compatibility with iOS and major Android platforms.

Experiments showed that embedding the AI entry in the framework was not prominent enough, whereas the pinch gesture was novel, attracted user interest, and received positive feedback.

1.2 Refined Guidance – Shaping User Pinch Mental Model

Initially a full‑page strong‑block guide was used, but it reduced information‑retrieval efficiency. The guidance was refined per page type (search results, image/dynamic feeds, third‑party landing pages, video detail pages, mini‑program pages), adjusting guide strength based on content readability and AI‑generated text quality.

Search and video detail pages: strong guidance due to high demand and high‑quality AI summaries.

Third‑party landing pages: medium guidance to improve recognition while leveraging good AI summaries.

Image/dynamic feeds and mini‑program pages: weak guidance to preserve browsing flow.

1.3 Full‑Page Container – Immersive Browsing

Using a full‑page container to display summary content maximizes screen usage, improves visual presentation, and creates an engaging, focused reading environment.

1.4 Dynamic Feedback – Enhancing Interaction

When users perform the pinch gesture, haptic feedback confirms the action, and they can freely drag the interface with two fingers, adding fun and making the interaction more intuitive.

2. Crafting an AI‑Intelligent Design Language

The goal is to create a smart, eye‑catching UI that highlights AI functionality and delivers a futuristic visual experience.

2.1 Intelligent Symbols – Enhancing Recognition

Lightweight “AI” logos and star symbols representing wisdom are integrated to allow quick identification of AI features.

2.2 Gradient Colors – Conveying Technological Feel

Through experimentation, a blue‑pink gradient was chosen to combine Baidu’s brand blue with a tech‑forward purple, delivering a modern, friendly AI aesthetic.

2.3 Expressive Animation – Communicating AI Concepts

Animations smooth the transition from “pinch trigger” to “content loading,” reducing user drop‑off. A typing‑effect displays summary text progressively, enhancing rhythm and efficiency.

2.4 Transition & Loading Effects

The container shrinks following the pinch gesture, matching user expectations. A breathing‑like scaling animation with synchronized ripple effects simulates AI “thinking,” providing an immersive experience.

Skeleton screens visualize AI data collection and analysis, setting clear expectations and lowering bounce rates.

3. Structured Layout to Promote Information Retrieval

3.1 Layout Structure

A hierarchical “summary‑detail” layout improves readability and guides users through information flow, with generous spacing and a minimalist gray background to highlight content.

3.2 Content Activation

Bottom sections can show “Related Recommendations” or “What Others Are Searching” to personalize the browsing experience.

3.3 Icon Matching Strategy

Icons are selected based on text themes and context to reinforce information, improve comprehension, and enhance visual appeal.

Conclusion

The “Pinch‑to‑Summarize” project demonstrates how integrating AI with design thinking can lower learning costs, reduce operational friction, provide responsive feedback, and shape a smart product perception through an elegant visual language and structured layout. Future Baidu App AI experiences will continue to pursue more natural interactions and lightweight visual cues, offering inspiration for designers undertaking AI‑driven product redesigns.