Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

IntentWeave: A Progressive Entry Ladder for Multi-Surface Browser Agents in Cloud Portals

Wanying Mo, Jijia Lai, Xiaoming Wang

Abstract

Browser agents built on LLMs can act in web interfaces, yet most remain confined to a single chat surface (e.g., a sidebar). This mismatch with real browsing can increase context-switching and reduce user control. We introduce \textbf{IntentWeave}, a design space of ten spatial paradigms for embedding agentic assistance across a browser, organized as a progressive entry ladder from micro-interventions to dedicated workspaces. We implement IntentWeave as a browser-extension prototype on the Alibaba Cloud website and compare three entry strategies in a within-subjects study (N=16). Workspace-heavy strategies reduced completion time but lowered perceived control; micro-only strategies preserved control but were often insufficient; a mixed sidecar approach achieved the highest satisfaction. We conclude with guidance for escalating and retreating agent surfaces without disrupting user agency.

IntentWeave: A Progressive Entry Ladder for Multi-Surface Browser Agents in Cloud Portals

Abstract

Browser agents built on LLMs can act in web interfaces, yet most remain confined to a single chat surface (e.g., a sidebar). This mismatch with real browsing can increase context-switching and reduce user control. We introduce \textbf{IntentWeave}, a design space of ten spatial paradigms for embedding agentic assistance across a browser, organized as a progressive entry ladder from micro-interventions to dedicated workspaces. We implement IntentWeave as a browser-extension prototype on the Alibaba Cloud website and compare three entry strategies in a within-subjects study (N=16). Workspace-heavy strategies reduced completion time but lowered perceived control; micro-only strategies preserved control but were often insufficient; a mixed sidecar approach achieved the highest satisfaction. We conclude with guidance for escalating and retreating agent surfaces without disrupting user agency.
Paper Structure (14 sections, 2 figures, 2 tables)

This paper contains 14 sections, 2 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Design Space: Ten Spatial Paradigms for Agentic Web Interaction
  3. Micro-interventions
  4. Embedded Surfaces
  5. Workspace Modes
  6. Case Study: IntentWeave on Alibaba Cloud
  7. Evaluation and Results
  8. User Study Specifications
  9. Experimental Design
  10. Procedure and Metrics
  11. Key findings
  12. Implications for design
  13. Limitations and Future Work
  14. Conclusion

Figures (2)

  • Figure 1: IntentWeave design space as a ten-paradigm pro gressive entry ladder
  • Figure 2: IntentWeave on Alibaba Cloud website. (A) Home Page: Pre-initialized assistant dialogue; returning users are greeted with a dialog preloaded with inferred intent or shortcuts. (B) Product Introduction Page: Context-aware widget prompt; AI detects relevant offerings and displays coupons or promos based on product context. (C) Document: Search-driven guidance and intent prediction; AI enhances search by predicting task intent and suggesting relevant documentation. (D) Solution Details Page: Inline highlights and expandable summary; users can highlight text to receive AI-generated summaries and glossary explanations. (E) Parameter Comparison Table: Hover-triggered tooltip analysis; hovering over model names shows tailored pros/cons based on user profile and use case. (F) Pricing Estimation Page: Contextual overlay interaction; AI overlay allows conversational queries about pricing, discounts, or suggestions. (G) Console Dashboard: Floating assistant icon and side panel; users can query for diagnostics or resource status via a quick-access floating UI.