Advanced · Systems

Agent Discovery (DAWN)

Visual diagram · (in preparation) · Math · (in preparation) · Worked example · 3 difficulty levels.

TL;DR. An emerging IETF framework — Discovery of Agents With Names — for finding out what an AI agent is, offers and whether it can be trusted before interacting.

Technical Definition

An emerging IETF framework — Discovery of Agents With Names — for finding out what an AI agent is, offers and whether it can be trusted before interacting.

How it works

DAWN addresses the discovery half of the agent web: an agent that needs to find another agent with specific capabilities first needs to know what it is, what it offers and whether it can be trusted — the same problem DNS and service discovery solved for networked machines, replayed for software that negotiates on your behalf. The IETF draft, published in early November 2026, standardizes the vocabulary: agent names resolvable to capability descriptions, credentials that attest who operates the agent, and metadata that lets a caller decide whether to proceed, challenge or refuse. Discovery is the missing front door for the agent economy — protocols like A2A assume you already know your counterparty; DAWN defines how you find it. Expect the terminology to settle as drafts progress, with 'agent discovery' as the durable term even if DAWN itself evolves.

Related Concepts

• Agent Experience (AX) — Designing websites, APIs and documentation so AI agents — not just people — can understand and use them easily.
• Agent Economy — An economy in which AI agents transact, negotiate and exchange services with businesses and with each other.
• Agent Identity — The verifiable answer to 'which agent is this?' — a stable, attestable identifier distinct from the human or service behind it.