Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

SoDA: An Efficient Interaction Paradigm for the Agentic Web

Zicai Cui, Zhouyuan Jian, Weiwen Liu, Weinan Zhang

TL;DR

The paper tackles data lock-in and cognitive overload in the emergent Agentic Web by introducing SoDA, an interaction paradigm built on orthogonal decoupling of storage, computation, and interaction. It equips the user with a Sovereign Memory Pod (SMP) and a Stateless Avatar Core that centralizes reasoning while preserving memory portability, governed by an A2A Intent-Permission Handshake with a dual-factor routing function $\Phi(S, R)$. Through high-fidelity simulation, the authors show substantial gains: cross-platform migration reduces tokens by approximately $27$–$35\%$, cognitive load drops by about $72\%$ (vs Strong RAG) or $88\%$ (vs manual), and information SNR improves, underscoring improved efficiency and privacy governance in a zero-trust environment. The work argues for a practical, decentralized, and privacy-preserving foundation for the Agentic Web, with UPDL enabling semantic interoperability and Burn-after-reading enforcing forward secrecy as data migrates across services.

Abstract

As the internet evolves from the mobile App-dominated Attention Economy to the Intent-Interconnection of the Agentic Web era, existing interaction modes fail to address the escalating challenges of data lock-in and cognitive overload. Addressing this, we defines a future-oriented user sovereignty interaction paradigm, aiming to realize a fundamental shift from killing time to saving time. Specifically, we argue that decoupling memory from application logic eliminates the structural basis of data lock-in, while shifting from explicit manual instruction to implicit intent alignment resolves cognitive overload by offloading execution complexity. This paradigm is implemented via the Sovereign Digital Avatar (SoDA), which employs an orthogonal decoupling design of storage, computation, and interaction. This establishes the architectural principle of data as a persistent asset, model as a transient tool, fundamentally breaking the platform monopoly on user memory. To support the operation of this new paradigm in zero-trust environments, we design an Intent-Permission Handshake Mechanism based on A2A protocols, utilizing dual-factor (Sensitivity Coefficient and Strictness Parameter) adaptive routing to achieve active risk governance. Empirical evaluation with a high-fidelity simulation environment indicates that this paradigm reduces token consumption by approximately 27-35\% during cross-platform service migration and complex task execution. Furthermore, in the orchestration of multi-modal complex tasks, it reduces user cognitive load by 72\% compared to standard Retrieval-Augmented Generation (RAG) architectures, by 88\% relative to manual workflows, while significantly boosting the Information Signal-to-Noise Ratio (SNR). These results demonstrate that the SoDA is the essential interaction infrastructure for building an efficient, low-friction, and decentralized Agentic Web.

SoDA: An Efficient Interaction Paradigm for the Agentic Web

TL;DR

The paper tackles data lock-in and cognitive overload in the emergent Agentic Web by introducing SoDA, an interaction paradigm built on orthogonal decoupling of storage, computation, and interaction. It equips the user with a Sovereign Memory Pod (SMP) and a Stateless Avatar Core that centralizes reasoning while preserving memory portability, governed by an A2A Intent-Permission Handshake with a dual-factor routing function . Through high-fidelity simulation, the authors show substantial gains: cross-platform migration reduces tokens by approximately –, cognitive load drops by about (vs Strong RAG) or (vs manual), and information SNR improves, underscoring improved efficiency and privacy governance in a zero-trust environment. The work argues for a practical, decentralized, and privacy-preserving foundation for the Agentic Web, with UPDL enabling semantic interoperability and Burn-after-reading enforcing forward secrecy as data migrates across services.

Abstract

As the internet evolves from the mobile App-dominated Attention Economy to the Intent-Interconnection of the Agentic Web era, existing interaction modes fail to address the escalating challenges of data lock-in and cognitive overload. Addressing this, we defines a future-oriented user sovereignty interaction paradigm, aiming to realize a fundamental shift from killing time to saving time. Specifically, we argue that decoupling memory from application logic eliminates the structural basis of data lock-in, while shifting from explicit manual instruction to implicit intent alignment resolves cognitive overload by offloading execution complexity. This paradigm is implemented via the Sovereign Digital Avatar (SoDA), which employs an orthogonal decoupling design of storage, computation, and interaction. This establishes the architectural principle of data as a persistent asset, model as a transient tool, fundamentally breaking the platform monopoly on user memory. To support the operation of this new paradigm in zero-trust environments, we design an Intent-Permission Handshake Mechanism based on A2A protocols, utilizing dual-factor (Sensitivity Coefficient and Strictness Parameter) adaptive routing to achieve active risk governance. Empirical evaluation with a high-fidelity simulation environment indicates that this paradigm reduces token consumption by approximately 27-35\% during cross-platform service migration and complex task execution. Furthermore, in the orchestration of multi-modal complex tasks, it reduces user cognitive load by 72\% compared to standard Retrieval-Augmented Generation (RAG) architectures, by 88\% relative to manual workflows, while significantly boosting the Information Signal-to-Noise Ratio (SNR). These results demonstrate that the SoDA is the essential interaction infrastructure for building an efficient, low-friction, and decentralized Agentic Web.
Paper Structure (17 sections, 4 figures, 3 tables)

This paper contains 17 sections, 4 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Agentic Web and Personal Agents
  4. Data Sovereignty and Decoupling
  5. HCI Evaluation Models
  6. System Architecture: A User Sovereignty Model for the Agentic Web
  7. Orthogonally Decoupled Three-Layer Architecture
  8. A2A-based Intent-Permission Handshake and Adaptive Routing
  9. Data Hot-Plugging and Semantic Interoperability
  10. Experiments and Evaluation
  11. Experimental Setup
  12. Task Benchmarks.
  13. Evaluation Metrics
  14. Comprehensive Evaluation across Multi-Modal and Cross-Domain Scenarios (RQ1)
  15. Evaluation of Data Decoupling and Cold Start Efficiency (RQ2)
  16. ...and 2 more sections

Figures (4)

  • Figure 1: The Human-as-Router Dilemma. Users face high cognitive load when manually coordinating context across isolated vertical agents, leading to severe information fragmentation.
  • Figure 2: The Orthogonally Decoupled Architecture of the SoDA. The system is structured into three functionally independent layers: (Top) Interaction Layer, which translates ambiguous user Intents into structured Directives for the Agentic Web and aggregates external Feedback into a low-cognitive-load Briefing; (Middle) Compute Layer, a stateless runtime that performs reasoning without data retention; and (Bottom) Storage Layer, the Sovereign Memory Pod (SMP) housing hybrid Knowledge Graph and Vector assets. A key innovation is the Hot-Plug mechanism, allowing the model to transiently access persistent data, ensuring user sovereignty by preventing data lock-in.
  • Figure 3: The Intent-Permission Handshake Mechanism. A rule engine acts as a privacy shield, dynamically approving or rejecting external agent requests based on user-defined policies and operational permissions.
  • Figure 4: Quantitative Evaluation of Interaction Efficiency. Comparative analysis of SNR, Cognitive Load ($L_{cog}$), and Interaction Friction ($\eta$) across three paradigms. The SoDA (Ours) demonstrates significant performance gains over Manual and General Agent.