Agentcities — The Global Agent Network Testbed

What Was Agentcities?

Agentcities was a global, collaborative research initiative dedicated to constructing an open network of online systems hosting diverse agent-based services. Launched in 2000 and active through 2005, it brought together researchers and institutions from more than 20 countries to build and operate the most ambitious distributed multi-agent testbed of its era.

The project operated at the intersection of several transformative technologies: FIPA agent standards, Semantic Web languages, Web Services infrastructure, and distributed GRID computing. Agentcities sought to prove that autonomous software agents could discover one another across open networks, negotiate, and cooperate to fulfil complex user goals — a vision that anticipated much of what we now call intelligent systems and service-oriented architecture.

At its core, Agentcities was a shared research testbed: a living network of FIPA-compliant agent platforms hosted by universities, research labs, and companies worldwide. Participating organizations connected their own platforms to the network, deployed agent services, and collaborated through working groups to solve technical challenges in agent interoperability, ontology alignment, and distributed service discovery.

Global agent network visualization — interconnected computing nodes across continents — The Agentcities network connected FIPA-compliant agent platforms across more than 20 countries, enabling autonomous agent discovery and service composition at global scale.

The initiative was coordinated by the Agentcities Task Force (ACTF) — an open body that organized meetings, managed working groups, and produced technical recommendations. European participation was funded through the EU 5th Framework Programme (Agentcities.NET project), while global involvement came through the broader research community and related projects.

Why It Mattered

In the early 2000s, the idea of truly interoperable software agents — systems that could autonomously discover services, negotiate terms, and coordinate across organizational boundaries — was largely theoretical. Agentcities made it practical. By creating a real, global testbed rather than a laboratory demonstration, the initiative generated invaluable insights into:

How agent platforms can discover one another in a decentralized network

The role of shared ontologies in enabling semantic interoperability

The intersection of agent technology with emerging Web Services and Semantic Web standards

Challenges of scaling multi-agent systems beyond controlled lab environments

Healthcare, rescue, and business process integration applications of agent technology

According to FIPA (Foundation for Intelligent Physical Agents), the standards developed and tested through Agentcities-era research remain foundational references for modern multi-agent systems design.

Explore the Archive

Dive into the research, standards, and community that made Agentcities a landmark initiative.

Network Architecture

How FIPA-compliant platforms connected globally: directory services, discovery mechanisms, and federated infrastructure.

Working Groups

Ten technical committees addressing architecture, ontology, web services, healthcare, security, rescue, and more.

Events

Four global Information Days — Sendai, Vancouver, Barcelona, Helsinki — plus the 2004 New York openNet Forum meeting.

FIPA Standards

The Foundation for Intelligent Physical Agents specifications: FIPA-ACL, agent management, interaction protocols.

MAS Concepts

Multi-agent systems theory: BDI architecture, emergence, agent autonomy, coordination and negotiation mechanisms.

Semantic Web & Agents

The convergence of agent technology with OWL ontologies, Semantic Web Services, and the openNet vision.

Research Projects

Agentcities.NET, Agentcities.RTD, national branches (ES, UK), and the openNet Forum evolution.

Applications

Real-world application domains: healthcare, emergency rescue, business process integration, and security.

A Legacy in Agent-Based Computing

The Agentcities project generated a rich body of technical recommendations, working group outputs, and conference proceedings. Its annual Information Days — iD1 in Sendai, iD2 in Vancouver, iD3 in Barcelona, and iD4 in Helsinki — brought together hundreds of researchers and culminated in live demonstrations of agent coordination across the global testbed.

The network eventually evolved into the openNet Forum, broadening its scope to encompass Semantic Web Services, GRID computing, and next-generation research testbed infrastructure. By 2005, the original testbed had wound down, but its intellectual legacy lives on in the agent-based computing and service-oriented architecture research communities.

This site serves as a reference archive of the Agentcities initiative — its mission, working groups, technical architecture, and the events that shaped early agent-based computing research.

Agentcities: The Global Agent Network