5 min read
This is an illustrative story. Dana is a composite based on common situations solo short-term rental hosts describe; the numbers are realistic examples, not figures from a single real business.
It is 11:40 on a Saturday night and Dana’s phone buzzes on the nightstand. A guest at her lakeside cabin cannot find the hot tub cover latch. Ten minutes later, a different guest at her second cabin asks whether the pancake place in town is open on Sundays. Neither message is urgent. Both get answered anyway, because Dana knows a slow reply can cost her a five-star review.
Dana runs two vacation rentals in the mountains about an hour from her home. She has no co-host and no property manager. Cleaning is contracted out; everything else, from listings to guest messages to reviews, is hers. By last spring she was spending about nine hours a week on guest messaging alone, much of it repeating answers she had typed a hundred times before.
Where the nine hours were going
Before changing anything, Dana spent one week tagging every guest message in a simple spreadsheet. The pattern was obvious by Wednesday:
- About 60% were questions already answered somewhere: Wi-Fi password, parking, check-in steps, the hot tub, trash day.
- About 25% were local questions: restaurants, hiking, grocery hours, where to rent kayaks.
- About 15% genuinely needed her: early check-in requests, a broken coffee maker, a complaint.
Only the last group needed Dana’s judgment. The first two needed accurate information, delivered quickly and warmly.
Step 1: One house manual per cabin, written once
Dana’s information was scattered across her listing, an old printed binder and her own memory. She sat down for two evenings and wrote one plain document per cabin, organized by the questions guests actually ask: arriving, parking, Wi-Fi, heating, the hot tub, trash and recycling, checkout, and what to do if something breaks.
She then added a separate “local guide” with her own tested recommendations, each with a note on seasonal hours and a reminder to check before going. Airbnb’s Resource Center for hosts has good examples of what guests expect in a house guide, and Dana borrowed its structure.
This step had nothing to do with AI, and it turned out to be the most valuable one. An AI assistant is only as good as the facts you give it.
Step 2: An AI project that drafts, and Dana who sends
Next, Dana created a project in Claude for each cabin and uploaded the house manual and local guide as its knowledge. Claude’s help pages explain how Projects keep files and instructions together, so every new chat starts with the right cabin’s details already loaded.
Her project instructions were short:
- Answer only from the uploaded documents. If the answer is not there, say “I’ll check with Dana and get right back to you.”
- Warm, brief, no more than four sentences.
- Never offer refunds, discounts, early check-in or late checkout. Flag those for Dana.
- Never include door codes or alarm codes.
When a message arrives, Dana pastes it in, reads the draft and sends it, usually with a small personal touch. Most replies now take her under a minute.
Step 3: Scheduled messages for the predictable moments
Many questions arrive at the same moments in every stay. Using her booking platform’s scheduled messaging feature, Dana set up four messages: a booking confirmation, a check-in guide sent two days before arrival, a first-morning check (“Is everything working well?”), and a checkout reminder. She drafted all four with AI help in one sitting, then rewrote them in her own voice.
The check-in guide alone cut arrival-day questions roughly in half, because guests had the answers before they asked.
Step 4: Reviews in one weekly batch
Instead of answering each review on her phone as it came in, Dana now handles them every Monday morning. Her AI project drafts a reply for each one, and she edits them so no two read alike. Our guide to answering reviews with AI in 15 minutes a week follows the same approach, and it works just as well for rental platforms.
What went wrong along the way
Not everything worked the first time, and the mistakes are the most useful part of this story.
- The invented brunch hours. In the first week, before Dana tightened the “answer only from the documents” rule, the assistant told a guest a cafe served brunch until 2 p.m. It closed at noon. The fix was the rule itself plus a “check hours before you go” line in every local recommendation. If you are new to this risk, our beginner’s guide to AI hallucinations explains why it happens.
- The too-generous draft. One draft offered a guest a free late checkout “as a courtesy”. Dana caught it before sending, and added the explicit “never offer” list above.
- Security details. Dana removed door codes from the documents entirely. Codes go out only through the platform’s own message, typed by her. We cover the wider privacy question in our comparison of private chat modes for client data.
The results after three months
- Guest messaging time: from about 9 hours a week to about 3.
- Late-night messages: down sharply, since most arrive answered by the check-in guide.
- Average response time: under 15 minutes during waking hours.
- Rating: steady, with several reviews mentioning “super clear instructions”.
The AI did not replace Dana’s judgment. It removed the typing. The 15% of messages that needed her still get her full attention, and now she has the time to give it.
How to copy this in one weekend
- Tag one week of guest messages into three groups: already answered, local, needs you.
- Write one house manual per property, organized by guest questions.
- Create an AI project per property with the manual and a short rule list, including what it must never offer.
- Draft four scheduled messages for booking, pre-arrival, first morning and checkout.
- Pick one weekly slot for reviews.
The same pattern works for any solo business with repeat questions, as our story about a solo dog groomer who cut no-shows shows. What is the question your customers ask you most often that you have never written down?
