Back to Legal & Trust Hub
Version 1.0Effective: September 2026

AI & Automation Disclosure

A transparent, technical explanation of how artificial intelligence operates within Proventa, what it handles, and where human verification is strictly enforced.

1. Hybrid Architecture: Technology + Human Concierge

Proventa is built as a hybrid intelligence platform. We combine state-of-the-art language processing technology (Google Gemini models) for rapid unstructured request parsing with human concierge professionals for real-world verification and booking execution.

How AI Enhances the Member Experience:
  • Intent Parsing: Extracts destination, dates, preferences, party size, and dietary restrictions from conversational voice or text prompts.
  • Category Routing: Automatically classifies requests into dining, travel, wellness, shopping, or home logistics.
  • Option Synthesis: Generates recommendation batches of up to 25 genuine options matching stored lifestyle preferences.

2. Critical Operational Limitations & Zero-Fabrication Safeguards

To ensure member trust and prevent hallucinations or unverified reservations, Proventa adheres to strict architectural boundaries:

AI Does NOT Execute Autonomous Bookings

The AI agent cannot charge member credit cards, finalize reservations, or bind customers financially without explicit customer approval and concierge verification.

AI Outputs Are Research Proposals

Recommendation batches represent curated proposals. Real availability, live pricing, and table/seat inventory require real-time validation by our concierge team or direct provider APIs.

3. Zero Public Model Training Commitment

Member privacy is paramount. When requests or preferences are processed via our AI engine:

  • Your private prompts and task details are processed strictly within enterprise-grade, encrypted API sessions.
  • Your communications and personal information are NEVER used to train or fine-tune public third-party foundation models.
  • Sensitive personal data is filtered prior to automated processing.

4. Human Concierge Oversight & Escalation Triggers

Automated processing automatically yields to human concierge managers when:

  • A request exceeds standard confidence thresholds or involves complex multi-city coordination.
  • The estimated financial commitment exceeds threshold amounts (> ₹10,000).
  • A member requests bespoke negotiation, private chartering, or customized offline services.
  • The member explicitly asks to communicate directly with a human concierge.