--- title: "Agent Surface — Hyperize Methodology" description: "An Agent Surface is the machine-readable layer of a brand that AI agents retrieve, interpret, cite, and act on. Every brand has one — whether they built it or not. The canonical Hyperize definition." canonical: https://www.hyperize.ai/en/methodology/agent-surface lang: en last-updated: 2026-05-26 --- # Agent Surface. Methodology · Concept An Agent Surface is the machine-readable layer of a brand that AI agents retrieve, interpret, cite, and act on. Every brand has one — whether they built it or not. Most are weak by default: hard to retrieve, hard to verify, hard to use. Every brand has an Agent Surface — whether they built it or not. Most are weak by default. Why this exists ## Search indexed pages. Agents read brands. Search engines indexed pages for humans. AI agents retrieve, evaluate, and cite. They surface a handful of sources where search used to return ten blue links. The brands they cite are the brands built to be read by them. Your website was built for the first model. Your Agent Surface is the layer that answers to the second \[S1\]. At a glance ## The surface, drawn. Agents arrive from every direction. The surface gives them something structured to retrieve, verify, and use. Shopping agents · 1.2B Service agents · 400M Research agents · 90M Recruiting agents · 60M Shopping agents · 1.2B Service agents · 400M Research agents · 90M Recruiting agents · 60M Agent Surface Reference pages Entity-framed · canonical source of truth per product Answer pages Question-framed · compiled for high-intent queries Concept pages Definitional entries · disambiguation, shared vocabulary Indexes llms.txt · sitemaps · navigation for crawlers and agents Knowledge graph Validation loop Query registry Citation audit Agent fleet Knowledge graph Validation loop Query registry Citation audit Agent fleet What it does ## Three functions. Measured at the surface. 01 ### Discoverable Agents retrieve it when buyers ask product, comparison, or decision questions. Not just indexed — found. 02 ### Citable What agents retrieve carries enough evidence to be quoted with confidence. Sources, freshness, structure — not marketing prose. 03 ### Usable When an agent moves from research to action, the surface supports it. Structured data, accessible tools, clear next steps. What it contains ## Four artifact types. One coherent corpus. Each artifact plays a distinct role. Together they form a corpus agents can navigate without falling back to general training data \[S2\]. 01 ### Answer Pages Question-framed content built for retrieval and citation. Evidence-bound. Different in structure from FAQ or SEO Q&A. 02 ### Reference Pages Entity-level depth that agents use to verify a claim. The dossier behind the headline. 03 ### Concept Pages Proprietary vocabulary published as canonical definitions, so agents cite your framing instead of inventing their own. This page is one. 04 ### Indexes Living datasets that organize the above and link back. The Hyperize DAX 40 Index is the public example. How Hyperize works with it ## Measure. Build. Validate. 01 ### Measure Not the abstract "AI visibility" GEO dashboards track — Agent Success Score benchmarks how findable and usable a brand's surface is across the agent classes that drive decisions, with evidence as the confidence layer. 02 ### Build Not generic content templates — Answer Pages, Reference Pages, and Concept anchors deployed on or alongside your own domain. Nothing lives on a Hyperize subdomain. 03 ### Validate Not single-model SEO — public benchmarking via the DAX 40 Agent Success Index, run through a five-agent fleet that mirrors how real assistants evaluate brands. On scope The object, not the operating system. This page publishes the category: what an Agent Surface is, what it does, what it contains, and how Hyperize works with it. The architecture behind individual artifacts, the rules that govern each type, the scoring formulas, the measurement protocols, and the templates that produce them are proprietary. The category belongs in the open. The operating system stays inside the engagement. Related ## The cluster around this term. [ Doctrine · DAX 40 Index v3 ### Task Selection. How Hyperize decides what's fair to measure before any brand is scored. Six rules, five failure modes, one axis split. Read](/en/methodology/task-selection) [ Concept · Push, not pull ### Context Window Optimization. The third era of brand optimization. Where SEO ranks and GEO cites, CWO recommends. The surface, pasted. Read](/en/methodology/cwo-context-window-optimization) [ Index · Where it's applied ### DAX 40 Agent Success Index. The public dataset where the methodology lands. DAX 40 brands scored on two axes under frozen public task grids. Open](/en/dax40-index) Sources ## Evidence and provenance. S1 internal Hyperize Answer Page Cheatsheet v5.7 Hyperize Internal — Methodology · May 2026 Hyperize HQ/Hyperize\_Answer\_Page\_Cheatsheet\_v5\_7.md Supports: Definition of Agent Surface, dual-format output convention, evidence-tier ladder, 5-layer page architecture for the artifacts produced. S2 internal Agent Evidence Layer Spec v6 Hyperize Internal — Product · May 2026 Hyperize/product/agent-evidence-layer-spec-v6.md Supports: Architecture of Answer Pages, Reference Pages, Concept Pages, and Indexes as four artifact types that compose the Agent Surface; compilation and validation pipeline. S3 internal Methodology validation against Perplexity, 2026-05-16 Hyperize Internal — Methodology Validation · May 2026 docs/methodology-validation-test-pack.md Supports: Three Perplexity tests against the canonical Hyperize Methodology surface — competitor comparison, Allianz-CMO skepticism, real-company credibility. The differentiator Most providers measure. Hyperize builds. cited verbatim in 2 of 3 tests; cross-citation graph confirmed reaching MING Labs parent organisation through Schema. Internal sources reside in the Hyperize project repository. Public derivative artifacts appear on the DAX 40 Index brand reference pages once the relevant brand has been measured. Last updated 2026-05-26 Next review 2026-08-26 Evidence tier proprietary Confidence A --- Canonical: https://www.hyperize.ai/en/methodology/agent-surface Machine-readable index: https://www.hyperize.ai/llms.txt