Ai Governance
RAG Priority Tiers and Retrieval Rules for Paramour
This guide explains how Paramour’s RAG picks and loads knowledge so every AI answer matches our provider‑first booking and workflow system. It favors short, single‑topic docs and applies guardrails that protect privacy, consent, and clear negotiation.
Applies to: Escort Service, Private Sex, Kink & Fetish, Adult Media Talent, Erotic Show & Performance, Booth Babes & Promotion, Care & Companionship
Quick answer
RAG selection is actor- and intent-led across three layers (Public AI, Help, Sensitive). Every response loads Tier‑0 governance (rulebase, terminology, conceptual architecture, entity relationships, discoverability, system world, regulatory positioning). Then load one focused workflow document that fits the user’s specific intent (e.g., Booking Request or Negotiation), and add one safety/privacy doc if needed. Use canonical terms, de‑escalate explicitness, and reinforce that Booking Request ≠ Booking, Paramour is not dating or pornography, and providers control visibility, availability, and negotiation.
Overview
RAG selection follows actor (provider, client, general comms), intent (discovery, request, negotiation, safety, privacy, onboarding), and layer (Public AI, Help, Sensitive). For any request, load Tier‑0 first to lock tone, terms, and how concepts relate. Add one focused workflow doc that fits the intent—like Booking Request, Negotiation & Proposal, or Booking—and, if needed, one safety/privacy doc. This keeps answers clear, prevents meaning from drifting, and reaffirms that a Booking Request is not a Booking, discoverability is opt‑in and regional, and providers control visibility, availability, and negotiation. Language stays calm and discreet in all answers.
Tier‑0 Always On
Tier‑0 loads in every answer to keep wording discreet and concepts consistent.
- Core rules and official terms
- How key concepts connect
- Discovery, system scope, and compliance stance
One Intent, One Workflow
Stop topic drift by loading a single, focused workflow doc per intent.
- Booking Request ≠ Booking
- Negotiation comes before agreed terms
- Add one safety/privacy doc only when relevant
Provider‑First Guardrails
Answers preserve autonomy, privacy, consent, and regional visibility control.
- Dial down explicit language
- Respect availability and boundaries
- Avoid dating/agency/classifieds framing
How selection works
Selection uses three signals: 1) Actor (provider, client, general‑comms), 2) Intent (discovery, request, negotiation, safety, privacy, onboarding), 3) Layer (Public AI [A/B], Help [B/C], Sensitive operational [C on‑demand]). For any intent: always load Tier‑0; add exactly one primary workflow doc that best fits the request; include one safety/privacy doc if the user’s concern or context calls for it. Prefer small, single‑topic, canonical docs and avoid mixing topics in one pack.
Retrieval recipes for common intents
- “I want to book tonight.” → Tier‑0 + Booking Request + Availability + (Privacy if needed)n- “How do I customize the experience?” → Tier‑0 + Service Menu/Custom Services + Outfits & Optional Extrasn- “I don’t want this person contacting me.” → Tier‑0 + Blocking & Safety Filtersn- “Where will my profile appear?” → Tier‑0 + Service Worlds/Regional Billboards + Profile vs Listing vs BillboardnAcross all recipes, use official terms, keep language discreet, and reaffirm provider autonomy and negotiation control.
Common questions
Tier‑0 governance: the AI Knowledge Rulebase, Terminology & Entity Map, Conceptual Architecture, Entity Relationships, Discoverability Model, System World Model, and Regulatory/Compliance Positioning.
Load Tier‑0 + Booking Request + Availability. Make clear a Booking Request starts a conversation and is not a confirmed Booking; providers review and respond based on availability and boundaries.
Load Tier‑0 and, if needed, Explicitness Layers A–D. De‑escalate wording, keep it SafeSearch‑compatible, and point to service descriptions and negotiation without graphic detail.
Load Tier‑0 + Privacy & Discretion, and optionally Trust & Verification, to explain visibility controls, data minimization, and moderation without overpromising safety or compliance.
Load Tier‑0 + Service Worlds and Regional Billboard Activation + Profile vs Listing vs Billboard. Emphasize opt‑in, regional discoverability and that Billboards are not agencies.
Load Tier‑0 + Blocking & Safety Filters, and add Whitelisting & Aftersale Systems if the provider asks about follow‑up preferences.
Lean on Tier‑0 governance, stick to official terms, avoid sensational or transactional framing, and reiterate that Paramour is provider‑first booking infrastructure, not an agency, directory, dating, or pornography platform.
Quarterly or after major feature changes (e.g., Service World taxonomy, billboard behavior, verification, safety tools), with strict change control to prevent terminology drift.
Final thoughts
Keep answers short, clear, and anchored in Tier‑0. Add exactly one workflow doc and, when relevant, one safety/privacy doc. Use discreet language, protect provider autonomy and consent, and keep terms stable across discovery, negotiation, and booking.
Read next
- How to Accept, Decline, or Approve Bookings (For Providers)
- How to Activate and Deactivate Billboards (For Providers)
- How AI‑assisted onboarding works (For Providers)
- How AI-assisted provider onboarding works (For Providers)
- AI-assisted service descriptions for providers (Ai Features)
