AI Transparency

Operhive uses AI to help you

When you interact with Apex, Nova or Nexus you are talking to an AI system. Here is how it works, why we use it, and how to reach a human.

Operhive’s three AI agents

Every automated reply you receive on WhatsApp, in the dashboard or on the website is generated by one of these agents. Each has a precise role and dedicated language models.

Apex

Host copilot in the dashboard and on WhatsApp

AI model: OpenAI gpt-5.4-mini on WhatsApp (fallback gpt-4.1-mini; vision gpt-5-mini, audio whisper-1) and gpt-5-mini in the web dashboard (fallback gpt-4.1-mini)
Purpose: Helps the host manage bookings, pricing, calendar, guests and operations through natural language.
Where you find it: Dashboard /apex and the property’s WhatsApp business number
How to opt out: The host can disable Apex from Settings > Automations; outbound messages require an explicit confirm-token (FSM propose → confirm → undo 30s).
When a human steps in: Every critical action (price edits, guest sends, cancellations) requires explicit human confirmation. Support: hello@operhive.com.

Nova

AI concierge for guests over WhatsApp

AI model: OpenAI gpt-5.4-mini on WhatsApp (fallback gpt-4.1-mini; vision gpt-5-mini, audio whisper-1)
Purpose: Replies to guests about availability, pricing, property info, check-in, activities and payments. Runs proactive journey messages.
Where you find it: Inbound WhatsApp conversations on the property’s number
How to opt out: The guest can ask for a human at any time: phrases like “human”, “staff”, “person” trigger immediate escalation to the host. The host can also disable AI from Settings > Automations.
When a human steps in: Any emergency (deterministic keywords IT/EN/DE/FR/ES) bypasses AI and pings the host. Out-of-scope requests are forwarded to the host.

Nexus

AI assistant on the public Operhive website

AI model: OpenAI gpt-5-nano (verified production runtime; no automatic fallback)
Purpose: Answers questions about Operhive (features, pricing, onboarding) for visitors and authenticated hosts on the website. No operational autonomy: information only and, on request, escalation to a human.
Where you find it: Chat widget on the public website operhive.com (public and authenticated modes)
How to opt out: Use is voluntary: you are not required to interact with the widget. You can close the chat at any time or request a human via the “Talk to us” tool.
When a human steps in: Nexus makes no decisions: it uses the “escalate” tool which opens a support ticket and forwards the request to a human (reply by email). No action on bookings, payments or data.

How to tell you are talking to an AI

Operhive labels every AI surface explicitly. If you can’t see one of these signs, please report it to privacy@operhive.com:

“AI Assistant” badge in the Apex header.
Disclaimer on the first turn of every new WhatsApp conversation with Nova.
Inline notice when the AI is not confident about the answer.
/ai-transparency page linked from the footer.

Your rights when interacting with Operhive AI

We are preparing Operhive for EU Reg. 2024/1689 (AI Act) by the applicable deadlines and process data under EU Reg. 2016/679 (GDPR). At any time you may:

Ask to speak with a human

On WhatsApp, type “human”, “staff”, “person”. On the site, click “Talk to us” in the widget. No AI can refuse the escalation.

Disable AI

The host can switch Apex and Nova off from Settings > Automations. Automated answers stop; only manual flow remains.

Export and delete the conversation

On request to privacy@operhive.com we provide a JSON export of AI interactions and erase them within 30 days (GDPR art. 17).

Review automated decisions

No Operhive AI takes legally significant decisions about a guest or host without explicit human confirmation (GDPR art. 22 by-design).

Regulatory references

This notice fulfils the following obligations of EU Regulation 2024/1689 (AI Act):

Art. 50(1) — “Providers and deployers of AI systems intended to interact directly with natural persons … shall ensure that natural persons … are informed that they are interacting with an AI system, unless this is obvious from the point of view of a reasonably well-informed, observant and circumspect natural person.”

Applicability deadline: 2 August 2026 (art. 113). Operhive is keeping disclosure, human escalation and audit trails aligned with the applicable requirements.

Penalty for non-compliance (art. 99(4)(a)): up to EUR 15,000,000 or 3% of worldwide annual turnover.