Toward Services Computing 3.0: Delegated Action and Agentic Service Relationships

When software agents choose and execute service actions, a correct task result does not by itself establish authorized action or fulfilled obligations. We advance a relationship-centered position for Services Computing 3.0: a principal’s bounded capacity to act is partially embodied in software, while authority, commitments, and evidence must remain inspectable as plans, carriers, and counterparties change. Building on services computing, service science, and multi-agent delegation, we distinguish four concepts—service-agent role, agentic service entity, agentic service, and agentic service system—and separate constitutive criteria from dependability requirements. A typed conceptual model and four operational rules expose action admission, commitment creation, revocation, and closure. Delegation contracts authorize action for a principal; collaboration contracts coordinate obligations among independently accountable parties. The position complements contemporary lifecycle, engineering, and agent-as-a-service perspectives. We operationalize the position through Agentic Service Value Engineering (ASVE) and the LuoShu platform, connecting enterprise application experience to value judgment, governed execution, and evidence review. This practice supports application feasibility and motivates specific relational controls; comparative benefits remain hypotheses. Constructed episodes clarify analytical consequences. The capability-to-fulfillment gap organizes five research lenses and an agenda for contract schemas, conformance traces, and comparative evaluation. The resulting research objects are reusable across implementations; the 3.0 label expresses a cumulative shift of emphasis.