| Internet-Draft | AI Agent Discovery for Communication Ser | November 2026 |
| Jadoon, et al. | Expires 6 May 2027 | [Page] |
This document presents a use case and problem statement for the initial discovery of AI agents supporting end-to-end communication services within an organization. The use case considers AI agents associated with different functions of the Internet protocol suite, including application, transport, Internet-layer, and related network-operation functions. These agents may share general capabilities but differ in their communication-related operational scopes and responsibilities. The document examines the minimum public discovery information needed to distinguish such agents within a known organizational discovery scope.¶
This Internet-Draft is submitted in full conformance with the provisions of BCP 78 and BCP 79.¶
Internet-Drafts are working documents of the Internet Engineering Task Force (IETF). Note that other groups may also distribute working documents as Internet-Drafts. The list of current Internet-Drafts is at https://datatracker.ietf.org/drafts/current/.¶
Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as "work in progress."¶
This Internet-Draft will expire on 6 May 2027.¶
Copyright (c) 2026 IETF Trust and the persons identified as the document authors. All rights reserved.¶
This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (https://trustee.ietf.org/license-info) in effect on the date of publication of this document. Please review these documents carefully, as they describe your rights and restrictions with respect to this document. Code Components extracted from this document must include Revised BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Revised BSD License.¶
The Internet protocol suite separates application, transport, Internet, and link-layer responsibilities [RFC1122] [RFC1123]. Each layer performs specific functions and interacts with other layers through established interfaces and protocols. This separation supports interoperability and allows the implementation and evolution of protocols independently of one another.¶
AI agents can augment these functions through adaptive decision-making and automation within their respective operational boundaries. For example, an application-associated agent may adapt media generation according to application requirements, while a transport-associated agent may assist with congestion control and transmission behavior. Similarly, an Internet-layer-associated agent may assist with routing and forwarding decisions, while agents associated with network operations may support packet scheduling, prioritization, and resource management.¶
In a layered agentic architecture, AI agents may operate independently within their functional boundaries and may cooperate with each other to support an end-to-end communication service. Each agent may use observations and interfaces relevant to its associated functions.¶
Research on cross-layer agentic networking illustrates cooperation among specialized agents [SANET], while work on intent-aware transport intelligence for immersive applications illustrates adaptation at a particular protocol layer [XR-TRANSPORT]. The architectural motivation for agentic augmentation of the layered protocol suite is further discussed in [LAYERED-AGENTS].¶
The ETSI Technology Radar identifies emerging ICT trends and assesses their potential implications for future standardization activities [ETSI-RADAR]. The deployment of specialized AI agents across communication functions raises interoperability considerations when agents need to cooperate to support shared service objectives. One prerequisite for such cooperation is the ability to discover agents exposing the required communication services and identify their associated protocol-layer responsibilities.¶
Before such interactions can occur, a Discovering Entity may need to obtain information about other AI agents that are discoverable within a given organizational scope. Similar high-level descriptions of communication-related services may refer to agents operating within different functional and operational boundaries.¶
This use case considers initial discovery within an organization, one of the deployment contexts identified by the DAWN Working Group [DAWN-CHARTER]. Subsequent capability exchange, agent selection, authorization, and operational coordination are outside the scope of this document.¶
This document uses the terminology defined in [DAWN-TERMS]. The following additional terms are defined for the purposes of this document.¶
In this document, a specialization of the DAWN term Function, referring to processing that supports an end-to-end communication service. Examples include application adaptation, transport optimization, routing, forwarding, and packet scheduling. A communication function describes the general service provided by an agent, rather than its detailed capabilities or internal implementation.¶
An architecture in which AI agents are associated with different communication functions across the Internet protocol suite and related network operations. These agents may cooperate to support end-to-end communication service objectives while operating within their respective functional responsibilities.¶
An AI agent whose advertised communication-related service is associated with one or more protocol-layer functions or related network operations.¶
An administrative or technological environment within which an AI agent provides its advertised communication function. An Operational Domain may be a subdivision of an organization's Discovery Scope. It describes the context of the agent's functional responsibility and does not, by itself, establish authorization to interact with the agent.¶
In this document, the unqualified term agent refers to an AI agent unless otherwise stated.¶
Consider an interactive augmented-reality (AR) application in an organization supporting maintenance in an industrial facility. A user device exchanges application traffic with a service through an access network and one or more intermediate networks. The access may be wired or wireless, and the service may be hosted locally or remotely. Application traffic and network load can affect the quality and continuity of the communication service.¶
The organization employs multiple agents to assist with different aspects of the communication service. An application-associated agent may adapt the resolution of AR content according to application requirements. A transport-associated agent may monitor transport performance and assist with transmission adaptation through interfaces permitted by the relevant transport implementation. Similarly, Internet-layer-associated agents may support routing and forwarding operations, while agents associated with network operations may assist with packet scheduling, prioritization, or resource management.¶
These agents may use different observations, interfaces, and operational controls. For example, an application-associated agent may observe media quality and application performance, while a transport-associated agent may observe round-trip time, loss, or transmission behavior. A forwarding-associated agent may use routing or topology information, whereas a scheduling-associated agent may rely on queue occupancy, traffic priority, or link conditions.¶
Although these agents contribute to the same end-to-end communication service, their operational responsibilities remain associated with different communication functions.¶
Within the organization, multiple AI agents may expose related high-level service descriptions. For example, a transport-associated agent and an Internet-layer-associated agent might both advertise services concerned with traffic optimization. However, the former may support transport adaptation, while the latter may support routing or forwarding operations. The same high-level description does not necessarily communicate these different protocol-layer associations.¶
Suppose a transport-associated AI agent needs to discover other AI agents exposing network-operation services within the organization's infrastructure. Acting as the Discovering Entity, it initiates discovery within the known organizational Discovery Scope. The discovery mechanism may return information about multiple AI resources that advertise related communication services.¶
For initial discovery, the relevant question is whether the returned public information is sufficient to distinguish the advertised communication functions and their protocol-layer associations, and to provide the reachability information necessary for subsequent interaction.¶
The DAWN requirements draft identifies entity type, offered services, reachability information, communication protocols, and relevant security parameters as components of Minimum Discoverable Information (MDI) [DAWN-REQ].¶
In the layered communication architecture described in Section 3, AI agents may expose similar high-level service descriptions while supporting different communication functions and protocol-layer responsibilities. For example, a general description such as traffic optimization may refer to an agent supporting transport adaptation, routing, or packet scheduling. These agents may have different operational scopes even though their advertised services are related.¶
The use case therefore raises the question of whether offered-service information, together with the Discovery Scope and other available discovery properties, is sufficient to distinguish the communication functions and protocol-layer responsibilities associated with discovered AI agents.¶
Problem Statement: What minimum public discovery information is needed to distinguish and contact AI agents exposing similar communication-related services but associated with different protocol-layer or operational-domain responsibilities, within a known organizational Discovery Scope?¶
This problem concerns the information made available during initial discovery rather than detailed capability assessment or agent selection. Depending on the discovery mechanism, the necessary distinctions might be expressed through offered-service descriptions, existing discovery properties, or additional discovery context. This document does not presume that new mandatory MDI fields or a particular discovery architecture are required.¶
This document describes an initial-discovery use case and does not specify a discovery protocol or security mechanism.¶
This document makes no requests of IANA.¶