The technical work of the Agentcities network was organized through working groups operating under the umbrella of the Agentcities Task Force. Each group owned a specific slice of the overall problem — from low-level message transport to high-level application domains such as healthcare and emergency response — and advanced its agenda through mailing lists, virtual meetings, draft specifications, and live experiments on the testbed. Together these ten committees turned a broad research vision into concrete recommendations and working prototypes.
Architecture WG
The Architecture Working Group owned the foundational question of how independent FIPA platforms could be federated into a single coherent network. It defined the models for cross-platform discovery, designed the hierarchy of domain-level and regional directories that allowed services to be located network-wide, and authored the platform-interoperability specifications that participants implemented. Crucially, the group validated its proposals against the live testbed, treating real deployment failures as design feedback rather than confining its work to whiteboard models.
Ontology WG
For agents to genuinely understand one another, they needed more than a shared message syntax — they needed shared meaning. The Ontology Working Group addressed this by developing common vocabularies and tackling the harder problems of ontology reuse and reconciliation across organizations that had each modeled their domains differently. Its work covered how to represent ontologies for use in agent messages, how to map between overlapping vocabularies, and how to let agents negotiate meaning when their underlying conceptual models did not perfectly align.
Communications WG
The Communications Working Group focused on the mechanics of moving messages reliably between agents across the network. Its remit included transport protocols, the reliability and ordering guarantees needed for dependable inter-platform messaging, and the definition of practical profiles of FIPA-ACL that constrained the full language to interoperable, implementable subsets. By clarifying how messages should be encoded, delivered, and acknowledged, the group reduced the ambiguity that otherwise caused conformant-but-incompatible implementations to fail when connected.
Web Services WG
As web services technologies gained momentum, the Web Services Working Group worked to bridge the agent and web-services worlds rather than treat them as rivals. It explored how SOAP and WSDL-based services could be integrated with FIPA agent platforms, allowing agents to invoke conventional web services and allowing web-services clients to reach agent-provided capabilities. This bridging work positioned Agentcities at the intersection of two communities that were independently converging on similar problems of distributed, loosely coupled service interaction.
ESOA WG (Extended Service-Oriented Architecture)
The Extended Service-Oriented Architecture Working Group looked beyond the specific protocols of traditional web services to the broader architectural principles of service orientation. It examined how the autonomy, goal-directed behavior, and negotiation capabilities of agents could enrich service-oriented systems — supporting dynamic, runtime composition of services rather than statically wired integrations. In doing so the group anticipated themes that would later become central to mainstream SOA and microservice thinking.
Healthcare WG
The Healthcare Working Group applied agent technology to one of the most demanding integration environments imaginable. Its work addressed clinical coordination across departments, patient monitoring, and the integration of multiple heterogeneous health information systems that rarely shared common data models. The group organized the iD3 Healthcare Workshop, bringing clinical and technical participants together to ground the network's capabilities in realistic medical scenarios where reliability, privacy, and interoperability all carried serious consequences.
Security WG
An open network in which agents from unfamiliar organizations interact continuously cannot rely on perimeter trust. The Security Working Group studied authentication, authorization, and trust in exactly this open setting. It investigated how agents could establish and reason about trust without prior relationships, how access to services could be controlled across organizational boundaries, and which security protocols were practical to deploy on a live, heterogeneous testbed without undermining the openness that made the network valuable in the first place.
Rescue WG
The Rescue Working Group targeted emergency-response coordination, a domain that stresses every property of a distributed agent system at once. Effective response demands cooperation among many independent organizations, dynamic allocation of scarce resources, and decision-making under acute time pressure and incomplete information. The group explored how agents could coordinate across organizational lines during a crisis, reallocating resources as situations evolved, and used these scenarios as a rigorous proving ground for the network's interoperability and discovery mechanisms.
BPI WG (Business Process Integration)
The Business Process Integration Working Group concentrated on inter-organizational workflow — the agent-mediated coordination of business processes that span enterprise boundaries. Its work included mediating between incompatible data formats, aligning processes that different organizations modeled in different ways, and enabling agents to manage long-running cross-enterprise transactions. By framing business integration as a coordination problem among autonomous parties, the group connected Agentcities directly to the commercial concerns of supply chains and B2B collaboration.
SDC WG (Service Directory and Composition)
The Service Directory and Composition Working Group addressed the dynamic end of service interaction: discovering distributed agent services at runtime and composing them into larger, value-added capabilities. Rather than assuming integrations were fixed in advance, the group studied how an agent could locate suitable services through the network's directories and assemble them on the fly to satisfy a goal. This emphasis on dynamic discovery and composition captured one of the most ambitious promises of the entire Agentcities vision.
The scholarly community that grew up around these working groups did not dissolve when the testbed wound down. Many of its members continued to publish and organize within the autonomous-agents research field, and the IFAAMAS (International Foundation for Autonomous Agents and Multiagent Systems) continues to organize AAMAS, the field's flagship conference, sustaining the research lineage that Agentcities helped to shape.