
AI Agent
An AI rostering assistant that automates the process of generating weekly rosters, managing staff and services on behalf of the Onsen Hot Pools team.
Key features
Onsen Hot Pools is a iconic tourism destination in Queenstown, offering multiple different kinds of services for travellers looking to be pampered with a view of the iconic Shotover river.
Behind the scences, rostering staff to these services is a time-consuming process that takes hours per week. An AI agent, with access to a custom algorthim, now automatically generates these saving Onsen 5 hours per week.
This AI agent for Onsens Rostering is powered by N8N, with access to a number of MCP servers and custom defined internal tools to manage Onsens rostering operations. The goal is to save 95% of the work for anyone managing rosters.
In the backend, Onsen has a Airtable database to manage staff, services and leave. Periodically,the Agent is called to generate future rosters which now only require a review from Onsen. This is powered by a custom annealing algoritihm exposed via an web-server, which the Agent can call as it needs to for rostering, forecasting etc.
In the front-end, Onsen can interact with this AI agent by a chat interface, getting it to complete common tasks related to staff management, but also generating rosters on demand formatted in a CSV they can then download to review.
This saves Onsen at minimum 5 hours a week per team (3 teams) making it a very effective automation built on good principles of human review points, and embedding into current workflows.
More work
Consulting engagement to optimize RAG pipeline, reducing costs while improving answer quality through better chunking, evaluation frameworks, and model selection.
An AI running coach that, powered by the brain of an ultrarunner, can coach any runner to a specific running goal.
A modern, AI-enabled full-stack web-application helping brands quote complex products & services with AI.
About us
A team of product engineers based in Queenstown, NZ. We work with you to understand the problem first, then build the right thing — not just the possible thing.
Book a consultation