Hospitality Guest Companion

The guest, first

A guest’s relationship with a boutique hotel is short, expensive, and emotional. They want three things from software: a smooth arrival (directions, check-in details, door codes), a low-friction way to ask for things (late checkout, a restaurant table, more towels), and local guidance that feels like a knowledgeable friend rather than a review-site dump. Anything beyond that is the hotel’s vanity, not the guest’s need.

Never a chatbot wall

The fastest way to cheapen a boutique stay is to put a bot between the guest and the house. The concept’s hard rule: guests always converse with named staff. Software disambiguates and routes — a towel request goes to housekeeping’s queue, a dinner question to the concierge — and AI works entirely on the staff side: translating messages, suggesting reply drafts, surfacing the guest’s context. Staff send every reply under their own name, edited or not. The guest experiences a responsive human; the staff experience a quieter shift.

Staff stay visibly human

Replies carry a name and role. Response expectations are honest (“housekeeping replies within 15 minutes, 7:00–22:00”) rather than the false immediacy a bot fakes. When the desk is closed, the app says so and takes a message — a hotel that admits its hours is more trustworthy than one pretending to be awake.

Adoption honesty

Nobody installs an app for a three-night stay. The concept is web-first: a link or QR code opens the full experience in the browser with no account creation — a signed per-stay link is the identity. The installable app exists for two audiences with a real reason to have it: returning guests, and properties whose guests average long stays. Every feature must work on the web version; the install is a convenience, never a gate. This single constraint kills half the feature ideas a hotel app usually accumulates, which is the point.

Local guidance, curated not scraped

Recommendations are the staff’s own list — the places they actually send people — structured with practical details (walking time, booking needed, quiet at lunch). No scraped reviews, no affiliate placement. AI helps staff keep the list current by flagging stale entries and drafting descriptions from staff notes, but nothing enters the guide without a person choosing it. A short list someone stands behind beats an infinite one nobody does.

Where AI stops

Beyond staff-side drafting and translation, AI is deliberately absent. No generative concierge persona, no automated upselling messages, no sentiment scoring of guests. Hotels earn loyalty through judgment and warmth; a model imitating either is a brand risk the concept refuses on principle.

Prototype scope and open questions

A credible prototype: the web guest surface (arrival, requests, guide), a single staff inbox with routing and AI-drafted replies, and per-stay link identity. The questions to answer: what share of guests engage via web link without any install prompt (adoption hypothesis); whether staff accept or fight the drafting layer; and whether request routing genuinely reduces front-desk interruptions or merely relocates them.