Beyond Application-Centered Computing: Toward an Autonomous Capability Ecosystem for Intent-Driven Internet Services

The Internet has traditionally been organized and accessed through applications, websites, platforms, and network-accessible services. Although this application-centered organization has enabled the rapid development of digital technologies, it also requires users and software systems to understand the boundaries of individual applications, identify appropriate service providers, and coordinate multiple digital resources when completing complex tasks. As generative artificial intelligence, autonomous agents, distributed computing, and interoperable digital services continue to develop, an alternative organizational perspective becomes increasingly relevant: the Internet could evolve toward an ecosystem in which computational capabilities, rather than complete applications, become the primary units of discovery, selection, composition, and execution. This paper proposes the concept of an Autonomous Capability Ecosystem (ACE) as a conceptual framework for organizing Internet resources around reusable, discoverable, interoperable, and dynamically composable digital capabilities. Instead of requiring users to identify and operate individual applications, an intent-driven system interprets the desired outcome, determines the capabilities required, discovers appropriate providers, and coordinates their execution across heterogeneous digital environments. Building upon the author's previously proposed frameworks, Generative AI and Search Engine Orchestration (GASEO) and Intent-Driven Interaction (IDI), this paper extends the investigation from the orchestration of specific Internet technologies and the interaction paradigm of human–Internet systems toward a broader question concerning the organizational architecture of the Internet itself. While GASEO examines the coordination of generative AI and search capabilities, and IDI establishes user intent as the primary basis of interaction, the present work investigates how the underlying Internet resources could be organized into a capability-oriented ecosystem. The paper introduces a conceptual distinction among application-centered, service-centered, and capability-centered computing. It further proposes a Capability-Oriented Internet Architecture (COIA), comprising capability representation, capability discovery, capability assessment, interoperability mediation, autonomous composition, distributed execution, outcome verification, and ecosystem governance. A capability is modeled independently of the application or service through which it is provided, allowing multiple providers to expose functionally comparable capabilities while retaining their independent implementations, ownership, operational policies, and execution environments. The proposed framework identifies several potential advantages, including reduced application dependence, cross-platform capability reuse, provider substitutability, more flexible task composition, improved resource utilization, reduced interaction complexity, and the possibility of developing an open and competitive capability ecosystem. It also examines fundamental challenges involving semantic interoperability, capability reliability, execution dependencies, security, privacy, accountability, economic incentives, and ecosystem governance. Importantly, this paper does not propose the elimination of applications, platforms, or service-oriented architectures. Instead, it explores a complementary organizational model in which applications and services become potential providers and containers of capabilities, while user intent and capability requirements provide a higher-level basis for Internet resource coordination. The work presents a theoretical framework and research propositions rather than claiming an experimentally validated replacement for existing Internet architectures.