What AI automation NZ businesses get from us, and the price
Most small businesses don't need a robot. They need the same ten jobs to stop landing on the owner at 9pm. Our AI automation work takes those jobs, builds them into your existing tools, and leaves a person in charge of anything that matters.
There are two ways in.
| Option | What you get | Price | Time |
|---|---|---|---|
| AI Operating System (AIOS) foundation build | Ops audit, a knowledge base of how your business runs, 3 to 5 core automations, connections to Gmail, Xero, Shopify and your booking system, team training, then 30 days of tuning | $5,000 | 4 to 6 weeks |
| AI app prototype sprint | A working prototype of a custom AI tool, such as a quoting assistant, an internal search tool or an AI receptionist. Credited to the production build if you go ahead | $3,500 | 3 weeks |
Jobs we automate for NZ SMEs
- Inbox triage. Every new email gets sorted into needs-you, can-wait and junk, with a draft reply ready for the ones that need an answer. You read, tweak and send.
- Quotes. Job details from an enquiry form or email become a draft quote using your own price list and wording. You check it before it goes.
- Xero invoice chasing. Overdue invoices are pulled from Xero each morning and a polite reminder is drafted for each one, matched to how late it is. You approve the batch, and the chasing gets done without you writing the same email again.
- Booking reminders. Clients get a reminder before their appointment and a rebooking nudge after, straight from your booking system.
- Daily reports. Yesterday's sales, bookings and anything unusual land on your phone before you start work, so you're not logging into three systems to find out how the day went.
- Review requests. Happy customers get a short request for a Google review a day or two after the job, sent from your own address.
- An AI receptionist. An assistant that answers the common questions on your website or messages after hours, takes the details and hands the conversation to a person. It doesn't pretend to be one.
A person approves anything that goes out or moves money
This is the rule every build follows. AI agents for business are good at reading, sorting and drafting. They shouldn't be trusted to email a client, issue a refund or change a price on their own. So anything that leaves the business or touches money stops at an approval step: a button in your inbox, a message on your phone, or a list you tick through once a day. Low-risk internal jobs, like filing an email or updating a report, run on their own. You decide where that line sits, and you can move it later.
Work we've built
UKIYO, Wānaka. UKIYO is a salon and haircare brand in Wānaka. Codie built its daily reporting out of Kitomba, the salon's booking and point-of-sale system, so the numbers arrive each morning without anyone running a report. He built a stock sync between Kitomba and the Shopify haircare store, so products sold in the salon and online are reconciled across both systems. And he built booking-gap alerts that flag empty spaces in the appointment book, so the team can fill them before the day arrives. These are the same kinds of systems we build for clients, running every day in a real business with real staff.
An accounting advisory. For RBK Advisory we built a management reporting system that pulls from Xero, Xero Practice Manager and Ignition, matches the same client across all three systems, and turns it into a directors' dashboard. It also takes their existing spreadsheet, so the dashboard fills before every connection is live.
An architectural ceilings and walls supplier. For MBS Architectural we set up a shared AI assistant loaded with their product information, spec sheets, supplier pricing and internal processes, so staff can get answers and draft quotes and documents without hunting through folders. We also built an overdue-invoice dashboard for their accounts team and a tested connection to their NetSuite system, and run weekly consulting sessions with the team.
How it works
- Ops audit (week 1). We sit with you and the people doing the work and list the jobs that eat the most time. We pick the 3 to 5 that are worth automating first. Some won't be, and we'll say so.
- Knowledge base (weeks 1 to 2). We write down how the business actually runs: prices, policies, tone of voice, who handles what. The automations read from this, which is why the drafts sound like you.
- Build and connect (weeks 2 to 4). We connect your tools (Gmail, Xero, Shopify, your booking system) and build the automations with approval steps in place. Nothing runs live until you've seen it work on real examples.
- Training (weeks 4 to 5). Your team learns how to approve, correct and switch things off. Short sessions, on Zoom or in person around Queenstown-Lakes.
- 30 days of tuning. We watch what the automations get right and wrong, and fix it. Most of the gains come in this month.
The foundation build takes 4 to 6 weeks end to end. If you'd rather test one idea first, the 3-week prototype sprint gets you a working tool you can try before committing to more.
Who it's for
AI automation for small business works best where the same admin repeats every day. We mostly work with:
- hair, beauty and barber salons
- trades and construction suppliers
- accounting, bookkeeping and advisory firms
- Shopify and online retail stores
- hospitality and tourism operators
- professional services, such as architects, consultants and agencies
If you're a team of 2 to 30 people and the owner or office manager is still the one copying data between systems, this is built for you.
NZ specifics: Privacy Act 2020 and your data
Business automation in NZ has to respect the Privacy Act 2020. In practice that means a few things we build in from the start:
- Only the data the job needs. An invoice-chasing automation needs a name, an amount and a due date. It doesn't need the rest of the client file, so it doesn't get it.
- Know where data goes. Most AI services process data offshore. The Act expects you to take care when personal information leaves New Zealand, so we tell you which services each automation uses and where they're based, and we use business accounts that don't train on your data.
- Access and correction. Clients can ask what you hold about them. Because everything is logged, you can answer.
- Breaches. Serious privacy breaches must be reported to the Privacy Commissioner. Tight permissions and approval steps make one far less likely.
We also connect to the tools NZ businesses already use, including Xero, Shopify, Kitomba and Gmail, so you're not paying for a new system to get started.
Pricing options
Pay in full, or ask us about spreading the cost over a few months. The AIOS foundation build is $5,000. The AI app prototype sprint is $3,500, and that amount is credited to the production build if you go ahead. You own what we build, and the knowledge base is yours to keep whatever happens next.
Not sure where to start? Book an AI 1:1 session ($399 for 60 minutes) and we'll map your first three automations together.