Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Collaborative Agentic AI Needs Interoperability Across Ecosystems

Rishi Sharma, Martijn de Vos, Pradyumna Chari, Ramesh Raskar, Anne-Marie Kermarrec

TL;DR

The paper argues that collaborative agentic AI faces systemic fragmentation unless interoperability is embedded from the start. It proposes Web of Agents, a minimal interoperability blueprint that reuses existing web standards across four building blocks—agent-to-agent messaging, interaction interoperability, state management, and discovery—to enable open, secure, and scalable cross-ecosystem collaboration. By analyzing current siloed solutions and illustrating how the four components interact, the authors advocate for minimal, extensible standards to prevent protocol wars and support decentralized, vendor-agnostic ecosystems. The work highlights security, open participation, and scalability as core benefits, while acknowledging broader trust, privacy, and safety challenges that must be addressed concurrently to realize practical, large-scale deployments.

Abstract

Collaborative agentic AI is projected to transform entire industries by enabling AI-powered agents to autonomously perceive, plan, and act within digital environments. Yet, current solutions in this field are all built in isolation, and we are rapidly heading toward a landscape of fragmented, incompatible ecosystems. In this position paper, we argue that interoperability, achieved by the adoption of minimal standards, is essential to ensure open, secure, web-scale, and widely-adopted agentic ecosystems. To this end, we devise a minimal architectural foundation for collaborative agentic AI, named Web of Agents, which is composed of four components: agent-to-agent messaging, interaction interoperability, state management, and agent discovery. Web of Agents adopts existing standards and reuses existing infrastructure where possible. With Web of Agents, we take the first but critical step toward interoperable agentic systems and offer a pragmatic path forward before ecosystem fragmentation becomes the norm.

Collaborative Agentic AI Needs Interoperability Across Ecosystems

TL;DR

The paper argues that collaborative agentic AI faces systemic fragmentation unless interoperability is embedded from the start. It proposes Web of Agents, a minimal interoperability blueprint that reuses existing web standards across four building blocks—agent-to-agent messaging, interaction interoperability, state management, and discovery—to enable open, secure, and scalable cross-ecosystem collaboration. By analyzing current siloed solutions and illustrating how the four components interact, the authors advocate for minimal, extensible standards to prevent protocol wars and support decentralized, vendor-agnostic ecosystems. The work highlights security, open participation, and scalability as core benefits, while acknowledging broader trust, privacy, and safety challenges that must be addressed concurrently to realize practical, large-scale deployments.

Abstract

Collaborative agentic AI is projected to transform entire industries by enabling AI-powered agents to autonomously perceive, plan, and act within digital environments. Yet, current solutions in this field are all built in isolation, and we are rapidly heading toward a landscape of fragmented, incompatible ecosystems. In this position paper, we argue that interoperability, achieved by the adoption of minimal standards, is essential to ensure open, secure, web-scale, and widely-adopted agentic ecosystems. To this end, we devise a minimal architectural foundation for collaborative agentic AI, named Web of Agents, which is composed of four components: agent-to-agent messaging, interaction interoperability, state management, and agent discovery. Web of Agents adopts existing standards and reuses existing infrastructure where possible. With Web of Agents, we take the first but critical step toward interoperable agentic systems and offer a pragmatic path forward before ecosystem fragmentation becomes the norm.

Paper Structure

This paper contains 15 sections, 3 figures, 2 tables.

Table of Contents

  1. Introduction
  2. From Generative AI to Collaborative Agentic AI
  3. The Alarming Trend Toward Incompatible Agentic Ecosystems
  4. Interoperability as Necessary Foundation for Agentic AI
  5. Security.
  6. Open Participation.
  7. Scalability.
  8. Architectural Foundations for an Interoperable Web of Agents
  9. Agent-to-Agent Messaging
  10. Interaction Interoperability
  11. State Management
  12. Discovery
  13. Putting it Together
  14. Alternative Views
  15. Final Remarks

Figures (3)

  • Figure 1: Left: Prominent protocols for collaborative agentic AI (A2A google2025a2a, ANP anp2025 and ACP acp2025). Agents by default cannot communicate with agents in other ecosystems, resulting in fragmentation (see \ref{['sec:incompatibility']}). Right: our proposed solution, Web of Agents. We advocate for ecosystem interoperability by encouraging practitioners to adopt existing minimal standards (see \ref{['sec:requirements']}).
  • Figure 2: Generative AI (left), agentic AI (middle), and collaborative agentic AI (right). This work provides a blueprint for interoperable collaborative agentic AI that leverages existing web protocols.
  • Figure 3: The blueprint of Web of Agents and its four building blocks.