Open source · by Joan Sanz

18 AI Agents
for the Hospitality
Industry.

A complete open-source repository of AI agent specifications for hotels. Each agent is a Markdown file you can drop into any LLM — Claude, ChatGPT, Kimi, Kilo Code, and beyond. We recommend Claude. Hoteltech SaaS platforms will follow. Just the spec, no lock-in.

⚡ Generate Your Agents Read the Vision →
18
AI Agents
5
Departments
4+
LLM Compatible
OS
Open source
The repository

All 18 agents

Each agent is a complete Markdown specification. Drop it into any LLM as a system prompt and you have a fully operational hospitality AI agent. Filter by department or explore all agents below.

Scope & Responsibilities

    Key Performance Indicators

      Escalation Rules

        How it works

        From spec to operational agent
        in 4 steps

        01
        Choose your agents
        Browse the repository and select the agents relevant to your hotel. Start with 1-2 agents, not all 18. The check-in agent and complaints agent are the highest-impact starting points.
        02
        Customize the .md file
        Open the Markdown file and replace the placeholder values: your hotel name, brand tone, compensation limits, escalation contacts, and the specific rules that apply to your property.
        03
        Load into any LLM
        Paste the .md content as a system prompt into any LLM. We recommend Claude — it handles long, structured specs exceptionally well. ChatGPT, Kimi, Kilo Code, and others work too. Hoteltech SaaS platforms will gradually support this format as the ecosystem matures.
        04
        Measure and improve
        Run the agent for 4 weeks. Measure the KPIs defined in the spec. Review escalation logs to identify gaps. Update the .md file based on what you learn. The spec gets smarter with every cycle.
        Works with any LLM · Hoteltech SaaS platforms will follow
        Claude — Recommended ★ Best for this
        ChatGPT
        Kimi
        Kilo Code
        + any LLM with system prompt

        The spec is LLM-agnostic by design. Today, drop the .md file into any LLM chat interface. As hoteltech SaaS platforms mature, they will natively support structured agent specs like this. The intelligence lives in the file — not in any platform.

        Live example

        The check-in agent in action

        This is what happens when the check-in agent receives a WhatsApp message from a guest arriving tomorrow. The agent reads the .md spec, checks the PMS context, and responds — in the hotel's voice, following escalation rules defined in the file.

        checkin-agent.md — Live session
        # System: [checkin-agent.md loaded as context] # Hotel: Grand Hotel Example | Room 412 | Arr: tomorrow GUEST_MSG "Hi, checking in tomorrow. Can I do early check-in at 10am?" AGENT_REASONING → spec rule: "Early check-in before 14:00 = subject to availability" → spec rule: "Gold member = complimentary early check-in if available" → check_loyalty_tier(guest_id='R412-SMITH') → GOLD → check_availability(date='tomorrow', time='10:00') → Room 312 available AGENT_RESPONSE "Good news, Mr. Smith! As a Gold member, we're happy to offer you early check-in at 10:00 AM complimentary. Room 312 will be ready for you — same floor, garden view. We'll send you a digital key at 9:45 AM. Is there anything you'd like prepared in your room?" → pms_update_checkin_time(reservation='R412', time='10:00') ✓ → notification_scheduled('9:45 AM digital key') ✓ █
        // Open source · works with any LLM · by Joan Sanz
        Ready to deploy
        your first AI agent?

        Use the generator to customize and download your agent package. Or read about the agentic future of the industry.

        ⚡ Generate My Agents The Agentic Future →