The Language of AI
Recommendation Infrastructure

AI Recommendation Infrastructure is the specialist-managed operating layer that connects the real capabilities of a hotel or portfolio to AI-mediated scenario demand. It governs the business's AI identity, evidence, Scenario Architecture, first-party machine-facing publication, Recommendation Intelligence, controlled intervention, retesting, synchronization, and protection. The category begins where simple visibility ends: a hotel can be known, indexed, cited, and frequently mentioned while remaining absent from commercially important AI-mediated decisions. External AI behaviour remains probabilistic and independent; Evidentity governs the business-facing infrastructure and measures the recommendation environment rather than claiming direct control over third-party models.

The Recommendation Economy is the market condition in which AI increasingly interprets traveler demand before conventional browsing has begun, compressing a broad universe of plausible hotels into a smaller Recommendation Set. Traditional search, OTAs and hotel websites remain important, but commercial selection can now occur before a hotel receives its first visit, enquiry or booking-engine session. The economic question therefore shifts from open discovery alone toward Recommendation Participation in the scenario-specific decisions that shape the shortlist.

AI-Mediated Demand is customer demand in which an AI system materially participates between the traveler's original intent and the businesses eventually considered. The underlying need already exists; what changes is the intermediary through which that need is interpreted, constrained, compared and allocated. The final booking may still occur through the hotel website, OTA, reservations team or booking engine, even though the commercially decisive narrowing of the market began inside an AI interface.

Pre-Click Demand is traveler demand already being interpreted, filtered and allocated before a hotel records its first measurable website visit. A traveler can ask an AI assistant for a shortlist, reject viable properties without opening their websites, and continue directly with one recommended option. Because no session, enquiry or abandoned booking is created for the omitted hotels, conventional funnel analytics remain blind to the loss. Pre-Click Demand describes the upstream commercial layer Recommendation Intelligence is designed to observe.

The Recommendation Economy is the overall market condition. The Scenario Economy is the structure through which that market divides into distinct decision opportunities defined by purpose, geography, constraints, evidence, timing, and commercial context.

Candidate Compression is the observable commercial process through which a broad market of plausible hotels becomes the much smaller Recommendation Set surfaced for one traveler situation. It is not a claim about one universal hidden algorithm or a single internal model mechanism. The commercial question is why some properties remain viable as the request becomes more specific while others disappear from consideration.

The group of businesses surfaced by an AI system for one particular request. It is scenario-specific, model-specific, and time-sensitive because a single material condition can change which candidates remain viable.

Recommendation Eligibility is the condition under which a hotel has enough real Scenario Fit and enough resolved identity, Operational Truth, evidence and decision-relevant clarity to remain a credible candidate for a specific AI-mediated request. It is not popularity, brand recognition or generic visibility. Eligibility depends on whether the hotel can genuinely satisfy the situation and whether that capability can be represented with enough coherence for the recommendation environment to evaluate it.

Recommendation Participation is the extent to which a hotel actually enters the AI-mediated decisions that matter commercially. It can include candidate-set inclusion, comparison with alternatives, scenario qualification, explicit recommendation, assignment of a commercial role, or routing toward an official booking, reservations or enquiry path. Participation is scenario-specific: a hotel can be strong for one traveler decision and absent from another. This makes it more commercially meaningful than generic AI visibility or a simple brand mention.

The practical ability of a business to remain commercially viable when AI narrows a broad market into a specific decision. It distinguishes being discoverable from remaining eligible when conditions become precise.

The practical boundary separating businesses that remain viable for a scenario from those that cease to fit as the request becomes more constrained. It reveals where real operating capability becomes commercially decisive.

Algorithmic Silence is the observable condition in which a hotel that appears relevant to a Recommendation Territory remains absent from meaningful AI-generated consideration. The term describes the outcome, not a presumed hidden model mechanism: the cause may lie in Scenario Fit, evidence, entity resolution, Operational Legibility, conflicting signals, stronger competitors or other conditions that Recommendation Intelligence must diagnose. Because the omission can occur before the traveler visits the property, the commercial loss may remain invisible to conventional analytics.

Scenario Demand is demand organized around the real traveler situation that needs to be resolved rather than a broad hotel category. A request for late arrival, a residential training programme, a quiet family stay or a dining-led city break creates a different practical market because the relevant operational conditions change. The scenario determines which capabilities matter, which evidence becomes material and which hotels remain viable.

The group of businesses genuinely competing to satisfy one defined customer situation. Membership changes as requirements change, so one business can compete in many different Scenario Markets.

A commercially meaningful area of AI-mediated demand defined by geography, scenario, customer constraints, business capability, and commercial purpose. It is broader than one prompt and more useful than a generic category.

Scenario Fit is the degree to which the actual hotel is appropriate for the operational requirements of a specific traveler request. It concerns the reality of the asset before its representation: whether the property can genuinely meet the relevant conditions around location, rooms, timing, group configuration, dining, access, privacy, transport or another material need. A hotel may have strong Scenario Fit while still lacking the evidence or representation needed to participate consistently in the recommendation environment.

Scenario Readiness is the degree to which genuine Scenario Fit has been translated into sufficient governed truth, evidence, operational clarity and machine-facing representation to support credible Recommendation Participation. It is the bridge between what the hotel can actually do and what can be established inside a recommendation decision. Scenario Readiness therefore depends on more than the physical asset: it also depends on the integrity of the identity, sources, publication and official handoff surrounding it.

The condition in which a business meets the full practical and informational requirements necessary to remain viable inside a specific AI-mediated request.

How far a business remains qualified as the customer adds increasingly specific requirements. It distinguishes superficial relevance from deeper recommendation readiness.

The point at which a business ceases to be the right answer. Clear boundaries are as important as capabilities because credible recommendation requires knowing where suitability stops.

The range of commercially meaningful scenarios in which a business has credible Recommendation Participation. It shows how much real commercial capability has been translated into observable AI-mediated market participation.

The set of AI-mediated scenarios for which the business has a genuine physical, operational, evidentiary, and commercial right to compete with the capability it already possesses.

The Addressable Recommendation Footprint is the complete map of recurring Recommendation Territories and Scenario Markets created by the real asset: the demand the hotel has a legitimate physical, operational, evidentiary and commercial right to serve. It is not a hypothetical aspiration or a generic growth wish. It defines the opportunity against which observed participation can be assessed.

The Observed Recommendation Footprint is the portion of the Addressable Recommendation Footprint in which independent AI systems currently include, compare, shortlist, recommend or otherwise treat the hotel as a viable candidate. It is an observed position within a defined testing universe, not a claim about total real-world AI market share.

A Recommendation Gap is the difference between a business's Addressable Recommendation Footprint and its Observed Recommendation Footprint. It exists where the hotel has a legitimate physical, operational, evidentiary and commercial right to compete for recurring scenario demand but observed AI-mediated participation does not adequately reflect that capability. The Recommendation Gap is an opportunity for diagnosis, not an automatic claim of lost revenue; its cause must still be established through Recommendation Intelligence.

A Scenario Competitor is the business receiving consideration for a specific traveler situation that the client hotel has a legitimate capability to serve. It is defined by demand allocation, not by visual similarity, star rating, proximity alone or a static legacy compset. The relevant competitor changes with the decision itself: the hotel that receives a late-arrival, family, meeting, wellness or residential-stay opportunity may not be the same property in every market.

The condition in which a competitor receives a recommendation opportunity that a client business was legitimately equipped to contest. It exposes the practical route through which AI-mediated demand moves elsewhere.

The sustained competitive force created when the same relevant businesses repeatedly receive scenario-specific recommendation opportunity instead of the client.

A Structural Loss is a recommendation loss in which another business genuinely offers a stronger fit for the traveler scenario. Structural losses are observed and understood, not artificially optimized against. Their purpose is to preserve intellectual discipline by distinguishing a true competitive advantage from a correctable infrastructure weakness.

An Addressable Loss is a recommendation loss in a market the hotel has a real right to compete for, where identity, evidence, representation, Scenario Architecture or commercial routing can still be strengthened. It is not assumed from absence alone; Recommendation Intelligence must establish that the loss is addressable rather than structural.

Demand Leakage is addressable AI-mediated opportunity that leaves the intended commercial path before the hotel records it in its own funnel. It can occur competitively, when another hotel captures a scenario the client was legitimately equipped to contest, or commercially, when interest in the hotel is routed through an OTA, directory or another unintended intermediary rather than the hotel's authoritative path. These are related but distinct forms of leakage: one concerns who receives the recommendation opportunity, the other concerns where an otherwise relevant opportunity is routed.

The Governed AI Identity is the complete AI-facing identity architecture of a hotel or portfolio: its operational reality, capabilities, evidence, policies, restrictions, commercial relationships, scenario-relevant attributes, official routes and the boundaries attached to material claims. The Canonical AI Profile is the central canonical model inside that identity. The AI Site is its public first-party machine-facing expression; Recommendation Intelligence observes external behaviour around it; Recommendation Control manages intervention and retesting; and Profile Protection maintains its durability over time.

The central profile within the Governed AI Identity. It is the business's canonical operating memory from which publication, scenario architecture, evidence governance, monitoring, and intervention are managed consistently.

The practical, decision-relevant reality of how a business actually works: the capabilities, policies, conditions, constraints, and relationships that determine whether a customer need can be served.

The approved and governed representation of Operational Truth that anchors the business as a coherent entity across machine-facing surfaces and recommendation operations.

The Recommendation Contract is the governance logic defining what the hotel's machine-facing infrastructure is authorized to represent as Stable Truth, which claims are conditional, who has authority over them, what remains unknown or explicitly absent, and where Live State or transactional systems must take over. It is not a legal contract with an AI provider. It is the operating boundary of the Governed AI Identity.

Claim Authority defines who or what has the legitimate right to establish a material fact and under which conditions it is safe for machine-facing publication. Provenance asks where information came from; authority asks who can properly establish it. A hotel or operator may hold authority over policies and operating details, a regulator over licensing, and a booking engine over live rates and availability. These sources do not carry the same authority for every claim.

The documented origin and supporting history of a governed fact or claim. Provenance connects the statement to the business, source, validation context, and evidence relationship behind it.

The explicit state attached to a material claim, such as confirmed, externally supported, self-stated, derived, provisional, planned, unknown, or not offered.

Bounded Truth is a factual representation that includes the conditions, limits and exclusions necessary for a capability to remain true in a real traveler decision. A capability may be valid only for specified rooms, dates, group sizes, routes or operating conditions. Bounded Truth prevents broad marketing language from overstating operational capability.

Declared Absence is a governed statement that a capability, condition or service is explicitly not available. It protects recommendation quality by preventing AI from inferring positive capability where none exists and by making the hotel's real boundary usable in scenario qualification.

Unknown State is the explicit classification of information that cannot yet be established. Unknown is not the same as No: it is not treated as confirmed capability, confirmed limitation or declared absence. It remains a governed open condition until the relevant authority, evidence or operational confirmation resolves it.

The difference between the business reality that exists and the portion of that reality that has been sufficiently investigated, confirmed, structured, and governed for recommendation use.

The ability for the business, its assets, capabilities, evidence, and official commercial relationships to be resolved as one coherent entity rather than a collection of conflicting digital fragments.

The degree to which a business's important facts, relationships, evidence, boundaries, and routes can be extracted and interpreted without unnecessary ambiguity.

Operational Legibility is the degree to which the real operating capability of a hotel is explicit enough for an AI system to understand how the property works inside a customer decision. Machine Legibility asks whether the machine can resolve and interpret the business; Operational Legibility asks whether it can understand the practical meaning of late arrival, group capacity, parking, meeting suitability, accessibility or dining logistics when those conditions determine the answer.

The specific hotel facts that materially determine whether a property can qualify for a scenario, such as room capacity, group configuration, arrival rules, parking, dining, accessibility, policy conditions, or the official booking route.

The depth and specificity of evidence supporting the facts that matter most to a scenario. Strong evidence density allows critical capability to remain intact through comparison and qualification.

A missing, weak, outdated, ambiguous, or disconnected evidence relationship that makes a real capability harder to establish inside a recommendation decision.

An AI Site is the dedicated first-party machine-facing environment through which a hotel publishes its Governed AI Identity for intelligent systems. Evidentity typically deploys this layer on hotel-controlled architecture such as ai.yourhotel.com, where identity, Operational Truth, Scenario Architecture, evidence, boundaries, provenance and official commercial routes can be expressed in a form designed specifically for machine interpretation. The AI Site does not replace the hotel's main website: the primary website remains the human-facing environment for brand, experience, persuasion, content and conversion, while the AI Site creates a parallel interface for AI-mediated decisions within the hotel's own digital estate. Where appropriate, the architecture can be reinforced through an additional machine-facing layer on the primary domain, such as yourhotel.com/llm.

The AI-Facing Publication Layer is the coordinated first-party publication system through which the Governed AI Identity becomes publicly accessible to intelligent systems. It can include the dedicated AI Site at ai.yourhotel.com, optional main-domain reinforcement such as yourhotel.com/llm, machine-readable endpoints, canonical entity relationships, structured representations, evidence references, metadata and official commercial handoff routes. The layer reduces unnecessary reconstruction while keeping publication under the hotel's own digital identity; it does not claim to force independent AI systems to prefer these surfaces.

A Machine-Readable Endpoint is a structured public route through which AI systems can retrieve selected governed facts, relationships, claim states and official handoff information with reduced interpretive friction. It is a component inside the AI-Facing Publication Layer, not the core product or a substitute for the Governed AI Identity, Canonical AI Profile or dedicated AI Site.

The degree to which relevant public sources represent the same core Operational Truth without material conflict. It creates a stable center of gravity for recommendation evaluation.

A material inconsistency between public representations of the business that creates ambiguity around identity, capability, policy, evidence, or commercial route.

Public descriptions, listings, pages, or data surfaces that exist without a governing relationship to the Canonical Truth and can therefore drift, conflict, or weaken interpretation.

Recommendation Intelligence is the observational system through which Evidentity tests how independent AI systems treat a hotel across defined commercial scenarios. It records inclusion, omission, qualification, comparison, competitive substitution, assigned role, proposition integrity, routing, Model Divergence, Recommendation Stability and movement from baseline. It does not read hidden model reasoning or proprietary confidence scores. It observes external behaviour through structured scenario testing and turns that observation into an operating view of recommendation position.

Repeated testing of defined, commercially meaningful customer decisions across leading AI systems. It replaces generic brand queries with the scenarios that actually allocate demand.

The calibrated record of observed recommendation behaviour across the agreed scenario universe before managed intervention begins. It establishes the comparable starting point for future movement.

Recommendation Delta is the difference between observed recommendation behaviour in two comparable testing states. It can involve participation, substitution, scenario coverage, routing, stability, model consistency or recommendation position. A Delta establishes measured movement inside the agreed testing universe; it does not claim exact causal attribution for every external model outcome.

A specific condition that suppresses legitimate recommendation participation, such as unresolved identity, missing evidence, weak scenario relationship, ambiguous policy, stale information, or poor commercial routing.

The disciplined analysis that determines why a business is absent, weakly qualified, misrepresented, or displaced in a scenario and whether the loss is structural, operational, evidentiary, representational, or commercial.

Recommendation Control is the managed discipline of changing the business-facing conditions that can legitimately be governed in response to observed recommendation performance. It operates across identity, evidence, Scenario Architecture, publication, source coherence, commercial handoff, intervention, retesting and protection. Recommendation Control does not mean control over independent AI models; it means governed control of the business-side infrastructure combined with disciplined measurement of external outcomes.

Baseline -> Diagnosis -> Intervention -> Republication -> Retest -> Current Position -> Protection. Monitoring alone is observation, and intervention without retesting is assumption. The Recommendation Control Loop joins both into a managed operating discipline: establish the opening position, diagnose an addressable gap, make a controlled change, republish the governed state, retest the same decision and protect the integrity of the improved infrastructure over time.

A documented change to the governed infrastructure addressing one diagnosed recommendation weakness. Interventions strengthen the business identity, evidence, scenario logic, publication, freshness, or commercial route.

Recommendation Recovery is the observed restoration of meaningful participation in a scenario where the hotel possesses the required real capability but had previously been excluded, displaced or unstable because of an addressable infrastructure gap. Recovery is established through comparable retesting rather than assumed from the intervention itself. It becomes more meaningful when the improved position persists over repeated testing.

The consistency with which the business remains included, qualified, or recommended across repeated testing of comparable scenarios over time.

Model Divergence is the difference in observed recommendation behaviour between independent AI systems for the same scenario. It is why Recommendation Intelligence observes multiple model tracks rather than relying on one favourable interface result. Divergence is an observable market condition, not an assertion that Evidentity can inspect the private reasoning of a model.

Scenario Volatility is the degree to which observed recommendation outcomes fluctuate across repeated tests of a defined scenario. It distinguishes durable Recommendation Participation from an isolated favourable output and makes clear why a single prompt response is not a sufficient measure of position.

The gradual divergence of public information from current Operational Truth as policies change, pages age, platforms update unevenly, and historic descriptions remain in circulation.

A material change in how AI systems interpret, qualify, or recommend a business over time, even when the underlying commercial asset has not fundamentally changed.

Truth Debt is the accumulated backlog of Operational Truth that has not been properly governed, updated, resolved or incorporated into the hotel's AI-facing identity. It is the knowledge layer of a wider debt model: when real operating knowledge remains unresolved, it cannot be represented reliably or used confidently in scenario decisions.

Signal Debt is the accumulated burden of weak, missing, duplicated, stale, contradictory or unmanaged public representations that make the hotel harder to interpret correctly. It is the representation layer of the debt model: governed knowledge may exist internally, but the public ecosystem still fails to communicate it coherently.

Eligibility Debt is the accumulated set of unresolved conditions suppressing participation in Recommendation Territories the hotel has a legitimate capability to serve. It is the commercial consequence of Truth Debt and Signal Debt: operational knowledge is insufficiently governed or represented, so scenario qualification remains weaker than the real asset warrants.

The risk that a business will be excluded, substituted, weakened, misrepresented, or inconsistently routed in AI-mediated discovery because its recommendation infrastructure or Scenario Fit is insufficiently strong.

The ability to maintain meaningful Recommendation Participation despite changes in models, prompts, sources, competitors, business facts, and the public information environment.

A deliberate change to recommendation-facing business truth that preserves authority, relationships, boundaries, evidence state, and cross-surface coherence.

The operating discipline through which the integrity, freshness, evidence state, Scenario Architecture, and machine-facing representation of the Governed AI Identity are maintained over time.

Handoff Authority is the hotel's right to define the official continuation path once stable recommendation truth reaches the point where live commercial action must begin. It establishes which booking engine, reservations team, enquiry route or other authorized destination should receive the traveler rather than allowing an uncontrolled third party to define the next step.

The Transaction Boundary is the line between Stable Truth and Live State. Governed identity explains durable context, while pricing, availability, inventory, live packages and booking conditions belong to transactional systems unless they are dynamically connected through an approved authority. The AI Site should route to those systems rather than pretend to own volatile commercial truth.

The extent to which a business is prepared to receive and convert intent created upstream through AI-mediated recommendation through a clear, current, and authoritative path.

The proportion of observed recommendation opportunities within a defined testing universe in which the business appears meaningfully. It is a controlled analytical measure, not a claim about total market share.

Recommendation Yield measures how effectively the real capabilities of a physical hotel asset are converted into observable Recommendation Participation. It connects rooms, dining, meeting infrastructure, wellness, location, accessibility, operational policies and service capability to capital productivity through the scenario demand the asset is equipped to serve. A low Recommendation Yield therefore describes an asset whose commercial capabilities materially exceed the AI-mediated participation currently being extracted from them. It is a commercial productivity concept, not a universal percentage unless a defined measurement model has been agreed.

The structural benefit created when a business maintains stronger Scenario Fit, identity integrity, evidence, operational legibility, scenario coverage, and operating discipline than relevant competitors.

The accumulated advantage created by mature Recommendation Infrastructure: governed identity, evidence depth, scenario coverage, intelligence history, maintained publication, intervention history, and the capacity to respond as the market changes.

The extent to which a business or physical asset is prepared for AI-mediated recommendation as an increasingly important demand interface, including governed capability, evidence, participation intelligence, and operating discipline.

The degree to which governed AI identity, documented capability, recommendation intelligence, evidence architecture, and operating discipline strengthen the asset's future-readiness, resilience, scalability, and capital-productivity story.

The governed representation of a multi-asset business above the individual-property level while preserving the distinct identity, role, capability, and scenario position of every asset.

The commercially meaningful position a specific property occupies within the portfolio's Addressable Recommendation Footprint, determined by the demand it is best equipped to serve.

The asset within a portfolio with the strongest legitimate claim to a recurring recommendation scenario. It identifies the natural primary destination for that demand without requiring exclusivity.

The combined range of Recommendation Territories a portfolio can legitimately serve across its assets. It becomes valuable when those roles are explicit enough for AI to direct demand to the correct property.

The condition in which AI-mediated demand is directed toward the portfolio asset best suited to the actual customer scenario, strengthening conversion, utilization, and portfolio productivity.

The condition in which one portfolio asset receives recommendation consideration that could more appropriately have gone to another sister asset. It can preserve demand or expose portfolio confusion.

The extent to which AI-mediated demand addressable by at least one asset remains inside the portfolio's recommendation universe rather than leaving for an external competitor.

Demand that one or more portfolio assets are genuinely capable of serving but that leaves the portfolio entirely because the right asset, relationship, coverage, or route is not represented clearly enough.

Portfolio Recommendation Advantage is the structural edge created when a hotel group manages AI identity, Asset Roles, Scenario Ownership, demand coverage, competitive position and internal routing as one coordinated recommendation system. Instead of optimizing individual properties independently, the group can preserve a wider share of addressable demand, direct opportunities toward the asset best equipped to convert them, and identify where demand is leaking externally despite the portfolio already possessing a suitable answer. In this form, Recommendation Infrastructure becomes a management capability of the operator itself rather than merely a property-level digital capability.

The degree to which a business can be found, recognized, mentioned, or surfaced within AI-mediated discovery. It is an important upstream condition, but not the final commercial outcome.

The family of practices intended to improve how brands, entities, sources, and content are discovered, interpreted, cited, or surfaced within generative AI systems. GEO strengthens discovery; Recommendation Infrastructure governs participation in scenario-specific commercial decisions.

Practices intended to improve how information is structured, retrieved, and presented by systems that answer questions directly. AEO improves answer presence and clarity; Recommendation Infrastructure extends into eligibility, substitution, intervention, retesting, and ongoing control.