Case study · 004 · Shacky
ShackyFrom manual guest replies to instant AI-assisted responses.
Shacky answered every guest enquiry by hand in Hostaway. A person logged in, read the message, and wrote the reply. Nucleotto connected Hostaway to a cabin-specific AI agent that answers common questions immediately, and kept people on review and escalation.
ShackyWhere they started
Shacky runs short-stay cabins and tiny homes across Australia, booked through Hostaway. Every guest enquiry was answered by hand.
Each message needed a person to log in, read the enquiry, find the right answer, and reply. Communication waited on someone being available.
What was getting in the way
What we saw could change
Most guest questions repeat. A cabin-specific knowledge base and clear rules can answer them accurately, with people on the questions that need judgement.
How the work ran
Inside the inbox: which guest questions arrive, how often, and how each is answered.
Shows how enquiries arrive in Hostaway and how each reply is written by hand.
Finds that most questions repeat, and can be answered from cabin-specific knowledge.
Question mix and knowledge map.
Build a controlled AI-assisted workflow.
An AI agent that answers incoming Hostaway messages from cabin-specific knowledge and clear rules.
Confirms the cabin knowledge and the rules for what the agent may answer.
Designs the agent, the knowledge base, the rules system, and the escalation path.
Agent and rules blueprint.
Approve the design.
A webhook from Hostaway to an AI agent on its own droplet, with knowledge, rules, and a feedback loop.
Provides cabin knowledge and real guest enquiries.
Builds the webhook, the agent, the knowledge base, the rules system, and the feedback mechanism.
Working guest-response agent.
Start the draft program.
A 4-week program of drafted responses, reviewed and fed back before wider automation.
Reviews drafted responses for quality and accuracy.
Drafts responses to real enquiries, captures feedback, and trains the agent.
4-week draft and feedback results.
Rely on it for common questions.
The agent answers common questions immediately. People handle review and escalation.
Owns cabin knowledge, review, and guest relationships.
Runs delivery, tunes the agent, and improves it from feedback.
Live guest-response workflow.
Extend into a broader CRM.
How the agent works
One webhook connects Hostaway to an AI agent on its own droplet. The agent answers from cabin-specific knowledge and rules, and people stay on review.
What changed
How we kept it controlled
Cabin-specific knowledge, clear rules, and human review hold the line. Six mechanisms keep it controlled.
A 4-week draft program ran before deeper automation.
People review agent responses and feed back on quality.
Answers come from each cabin's own knowledge base.
Clear rules govern what the agent can answer.
Questions outside knowledge or rules go to a person.
Feedback trains and improves the agent over time.
What improved
Where it can go next
Shacky turned repetitive guest messaging into a controlled AI-assisted workflow.
Nucleotto connected Hostaway to a cabin-specific agent, and kept people on review.