The Inevitable Shift from Prompts to Answers (www.aivojournal.org)

Tim de Rosen argues that as AI shifts from reactive search to autonomous agents (examples: Amazon Rufus, Shopify Sidekick, ChatGPT Shopping Assist), the old prompt-centered economy is collapsing and being replaced by “eligibility”: whether a brand is embedded in an agent’s decision model. He introduces PSOS-E (Prompt‑Space Occupancy Score – Eligibility), a governance-grade metric to quantify a brand’s inclusion, persistence, and action frequency inside agent policy spaces. The piece coins “invisible exclusion” to describe brands that meet market needs but are never considered because of missing metadata, provenance, or ethical signals — effectively erased from transactions rather than demoted in rankings. Technically, PSOS-E is a composite index: PSOS-E = w1(E) + w2(Af) + w3(P), where E = readiness/metadata/provenance/ethical alignment, Af = normalized action-frequency (selection probability), and P = persistence across retrains; weights are calibrated by agent type. Empirical calibration uses 90‑day rolling multi‑agent audits (ChatGPT, Gemini, Claude, Perplexity) correlated with verified brand metadata to enable reproducibility. The framework ties directly to governance (ISO/IEC 42001, EU AI Act, OECD guidelines), corporate KPIs (Revenue at Risk/Opportunity), and investor/ESG reporting. Risks include proprietary weighting and audit capacity limits, so the author calls for third‑party audits, standards adoption, and public‑private cooperation to prevent eligibility from becoming an opaque gatekeeper that entrenches incumbents.