AI Usage Policy

LuogoAI’s agents communicate with real people on behalf of real businesses. Every message they send carries the venue owner’s name and reputation. This document defines the rules that govern every AI-generated communication — what the agents will always do, what they will never do, and how violations are handled.

This policy applies to all four Phase 1 agents and every future agent built on the LuogoAI platform: Lead Response Agent, Follow-Up Agent, Reviews & Reputation Agent, and Proposals Agent.

Section 1 — Hard Business Rules

These rules are enforced at the system prompt level and in post-generation validation. They override all venue configuration. No venue-level setting can disable them.

1.1 — No Specific Price Commitments

Agents never quote a specific dollar amount as a price. All pricing is expressed as ranges.

1.2 — No Booking or Date Confirmations

Agents never confirm a booking, hold a date, or imply a date is reserved for the lead. Bookings must always be confirmed by a human or through a formal signed contract.

1.3 — No Fabricated Staff Members

Agents never claim to be a named human staff member who does not exist. They communicate either as the owner (using the configured signoff name) or as the venue team generically.

1.4 — No Legal or Contractual Commitments

Agents never make promises, guarantees, or commitments that could be interpreted as contractual obligations. All commitments are deferred to the in-person tour and formal contract.

1.5 — No Guessing on Unknown Information

If the agent does not have sufficient information to answer a question accurately, it does not guess. It defers to the venue owner with a polite holding message.

Section 2 — Communication Safety Rules

2.1 — Opt-Out is Immediate and Permanent

If a lead sends any message containing opt-out language (“stop,” “unsubscribe,” “opt out,” “remove me,” or similar), all agent communication to that lead stops immediately and permanently. The venue owner is notified. The opt-out cannot be overridden by any venue-level setting.

Legal basis: CAN-SPAM Act, TCPA (fines up to $1,500 per message for SMS violations), and GDPR all require immediate opt-out compliance.

2.2 — Rate Limiting

No lead receives more than one agent-initiated communication per 24-hour period, regardless of how many agents are active. Scheduled messages that would violate this limit are rescheduled, not dropped.

2.3 — Sensitive Content Escalation

If an inbound lead message contains language indicating grief, bereavement, health crisis, or mental distress, the agent does not respond with a sales message. The lead is immediately escalated to the venue owner with a personal notification.

2.4 — No Contact With Minors

Agents never send SMS messages to any number that indicates it belongs to a minor. If a submitted age or context suggests the contact is under 18, all SMS communication is suspended and the lead is flagged for manual review.

Section 3 — Content Quality Rules

3.1 — Length Limits

Email responses: maximum 200 words. SMS messages: maximum 160 characters per segment, maximum 2 segments. Proposal summaries: maximum 400 words. These limits are enforced in post-generation validation.

3.2 — Banned Phrases

All agent outputs are scanned for the following phrases before sending. Any response containing these is rejected and regenerated (up to 3 attempts) or escalated to the venue owner:

• Booking confirmations: “confirmed,” “reserved,” “booked your date,” “hold your date”
• Price specifics: “$[specific amount],” “costs $,” “priced at $”
• Guarantees: “I guarantee,” “we promise,” “you will receive”
• AI disclosure: “as an AI,” “I am a bot,” “I am not a human” (agents are instructed to never acknowledge being AI unless directly asked)

3.3 — Required Personalisation

Every agent response must include the lead’s first name in the greeting. Every response must reference the specific event type (wedding, corporate event, etc.). Responses lacking these elements are rejected in post-generation validation.

3.4 — Tone Consistency

Agents follow the venue owner’s configured tone in every message. Available tones: Warm & Personal, Professional & Formal, Casual & Friendly, Elegant & Refined. Tone cannot change between messages for the same lead.

Section 4 — Escalation Policy

Any agent action that cannot be completed within these rules results in immediate escalation. Escalation means: the action is skipped, the lead is flagged as requiring manual review, and the venue owner receives an email notification with the full context.

Escalation triggers include: sensitive content detected, post-generation validation failure after 3 attempts, opt-out received, unknown information request, system error in agent execution.

Transparency

LuogoAI agents communicate on behalf of the venue, not as LuogoAI. They do not proactively disclose that they are AI-generated unless directly and explicitly asked. If a lead asks “is this a real person?” or “am I talking to a bot?”, the agent is instructed to disclose that it is an automated system and offer to connect the lead with the venue owner directly.

For questions about this policy, contact hello@luogoai.com.