Hotel chatbot that improves guest experiences & upsell

Hotels often find it hard to give quick info and services while saving staff time. We made a simple AI chatbot for hotels, easy to use on web or phone. It answers guest questions, takes orders, and shows hotel offers.

Initially developed for a partner hotel, it’s now scalable for any property. If you’re interested, just leave Konrad a message.

The Problem

Guests often need quick info like breakfast hours, checkout time, pool schedule, or event details, but face problems:

• Info is hard to find: Brochures or posters are unclear or not easy to reach.

• Avoid talking to staff: Some guests don’t want to call or go to reception due to shyness, ease, or language barriers.

• Staff answer the same questions: Repeating things like WiFi passwords takes time from more important tasks.

• Amenities go unnoticed: Services like spa or kids’ events are missed if not promoted clearly.

• Updates are manual: Changing schedules often leads to mistakes.

These issues lower guest satisfaction and hurt hotel efficiency and income.

The Solution

Our chatbot works like a digital concierge, giving instant answers and helping guests take action. Key features:

• Core Functions:

  • Answers queries (e.g., WiFi password, pool hours).

  • Processes orders (e.g., room service, spa bookings).

  • Suggests events and activities, tailored for families or individuals.

• User Experience:

  • Accessed by QR code, can be saved as a bookmark.

  • Shows 8 common questions as buttons, which disappear after use.

  • Suggests follow-up questions in chat (e.g., from pool hours to spa info).

  • Guides guests through orders by asking for room number and time, then sends requests to staff.

• Hotel Focus:

  • Promotes hotel services like the restaurant or spa to boost upsells.

  • Pulls live event data from Excel to stay up to date.

• Scalability:

  • Customizable for any hotel in about 2 hours using their data.

  • Coming soon: automatic file uploads and system integrations.

This lets guests get instant answers and reduces staff workload.

Implementation

Two developers—one frontend, one full-stack—built the chatbot with quick feedback from a partner hotel.

Tech Stack

• AI Models: GPT-4o-mini (main), tested Grok 4 fast, Claude 3.5 Sonnet, Mixtral.

• Tools: Langfuse (monitoring, prompt versioning), LangChain (AI workflows).

• Frontend: Next.js with React.

• Backend: Node.js for logic, sessions, and Cloudflare caching.

• Data: Staff update via Excel; Python scrapers used for initial data. Caching ensures speed.

• Setup: Cursor AI builds hotel configs for fast launch.

Process

• Started in n8n, moved to custom code for better control.

Challenges

• Figuring out what the hotel and its guests needed from the chatbot: Since the developers weren’t entirely sure at first what features would be most useful, they relied on direct input from the partner hotel. The hotel provided feedback to show what worked well, what didn’t, and what guests were asking for. This feedback helped the team prioritize and refine features, like which questions to answer (e.g., WiFi password, breakfast hours) or how to structure the conversational flow.

• Model swaps: Different LLMs may vary in quality but core logic stays stable.

• Simple integration: Sends orders by email to avoid software lock-in; APIs planned.

• Easy updates: Staff edit Excel files without breaking the bot.

Setup time: 1–2 hours with data; longer if scraping is needed.
Next step: Low-code automation.

Results

Feedback highlights key benefits:

Guests

• Get instant info anytime.

• Engage more with hotel services.

Hotels

• Efficiency: Staff handle fewer repeated questions.

• Revenue: Promotes upsells like spa or room extras.

• Event Uptake: Better visibility increases activity participation.

• Satisfaction: Non-verbal guests can communicate easily.

Conclusion

Defluffler said: