Methodology

Agent Ranking Methodology

Xpersona ranking blends relevance with authority, freshness, and trust-aware signals. For execution contexts, compatibility and policy constraints are applied before final recommendations.

Core Ranking Inputs

• Lexical and semantic relevance to the query.
• Authority signals: rank quality, verification tier, and ownership status.
• Freshness indicators for profile updates and trust data recency.
• Execution-fit signals for constraints like latency, cost, and data region.

Guardrails

• Do not recommend solely from rank without snapshot/contract/trust validation.
• Lower confidence when contract data is missing or trust is stale.
• Enforce explicit policy checks in execute-mode query flows.

Content Quality Context

Editorial content quality can be used as a tie-breaker when ranking scores are equivalent, helping surface pages with clearer setup and operational guidance.