The AI that refuses to make things up.
Built on Claude, Anthropic's safety first models. Our router automatically picks Claude Haiku for fast FAQ replies, escalates to Claude Sonnet for complex conversations, and reaches for Claude Opus on the edge cases where every word matters. You never pick a model; we handle every routing decision for you.
What you get with AI Concierge.
Powered by Claude, full stop
No swapping between chatbots from different vendors. Just Claude: Haiku for speed, Sonnet for complexity, and Opus (on the Pro tier) for legal questions, complaints, and high value bookings.
Smart routing, no model picker
Qontaktly chooses the right Claude model for every guest message automatically. Simple check in questions run on Haiku in milliseconds; longer refund threads escalate to Sonnet; Pro tier conversations with VIP flags route to Opus.
Knowledge base, built on yours
Upload your FAQ, house rules, menu, and local guides. Claude only answers from your content. If it does not know, it says so and hands off to a human; never invents an answer.
Multilingual by default
English, Bahasa Indonesia, Thai, Tagalog, Mandarin, Malay, Vietnamese. Claude replies in whatever language the guest wrote in.
Human handoff on every complaint
Automatic escalation when the guest mentions refund, complaint, accessibility, or security. A human team member picks up the conversation with full context and history.
Spend caps with live reply estimates
Each tier includes a monthly compute budget plus a live reply estimate that recalculates daily from your property's actual routing mix. You always know how many replies are left in the tank.
Who uses it.
- Answering 'what time is check in?' at 3 AM without waking up staff
- Sending parking and airport pickup directions in 7 languages automatically
- Escalating a guest complaint to the duty manager with full conversation history
Questions, answered.
Why only Claude? Why not GPT or Gemini too?
Because hospitality AI that makes up answers loses you guests. Claude's constitutional training means it is the model most likely to say 'I do not know' instead of fabricating a check in time or inventing an amenity. That reliability is worth more than the ability to toggle between providers. And because we focus on Claude only, we pour all of our prompt engineering and routing work into getting one family of models perfectly right for hospitality.
What is in each AI tier?
AI Starter ($399/mo) includes $50 of Claude compute, roughly 10,000 guest replies a month using Haiku and Sonnet. AI Growth ($599/mo) doubles the compute to $100, around 20,000 replies. AI Pro ($899/mo) includes $300 of Claude compute, ~60,000+ replies, plus access to Claude Opus for sensitive conversations and per property prompt fine tuning.
What does 'spend cap with reply estimate' mean?
Each tier has a Claude compute budget in USD. Your dashboard shows a live reply estimate calculated from your property's actual routing mix, recalculated daily. No surprises at month end; if you are running low, we tell you in advance so you can upgrade or throttle.
Does the AI invent answers?
No. Claude only replies from your uploaded knowledge base and live reservation context. If the question is outside that scope, it says so and hands the conversation to a human.
Can I turn off the AI on specific channels?
Yes. Enable the AI on WhatsApp, email, web chat, and the booking engine independently. Many properties start with WhatsApp only, then expand once they trust it.