AI development and automation for New Zealand businesses

Where AI helps a smaller business first

Most New Zealand firms are small or mid-sized, with few people covering many jobs. That is a good fit for narrow automation rather than large AI programmes. The best first project is usually a repeated task with clear inputs and a person who can check the result.

Privacy Act 2020 and customer data

The Privacy Act 2020 applies when a chatbot or automation handles personal information. Its information privacy principles cover why you collect data, how you store it, how long you keep it and who can see it. The Office of the Privacy Commissioner has published guidance on how those principles apply to AI tools, so read it alongside this page.

• Tell people when they are talking to a bot, and say what it does with what they type.
• Do not send more personal information to a model than the task needs. Strip names and account numbers where you can.
• Check the model provider's data terms, including whether inputs are used for training and where they are processed.
• Information privacy principle 12 limits disclosure of personal information outside New Zealand. Model APIs and our team in India both sit outside the country, so we record the arrangement in writing. Confirm the position with your lawyer.
• Agree a plan for a notifiable privacy breach before launch, including who calls the Commissioner.

Human review and honest claims

The Fair Trading Act 1986 prohibits misleading conduct, which includes what a chatbot tells a customer about price, stock or terms. A bot that invents a refund promise is still your promise in the customer's eyes. We keep answers grounded in documents you approve, make the bot say when it does not know, and route sensitive questions to a person.

Good scoping also keeps costs down. We list what the tool may do, what it must never do and who is told when it fails. A bot that can read an order status but cannot issue a refund is easier to trust than one with open access. Logs record each answer, so you can see what customers were told and fix gaps in your source documents.

Email and SMS automations are covered by the Unsolicited Electronic Messages Act 2007. Messages that sell something need consent, clear sender details and an unsubscribe route. We build those into any follow-up sequence. Our AI use policy template is a starting point for staff rules.

How a project runs from India

We work remotely, with no office in New Zealand. New Zealand is 6.5 hours ahead of India in winter and 7.5 hours ahead during daylight saving, so a 3 pm Wellington call is 7:30 or 8:30 in the morning in Pune. One lead owns your project, and you see a working version early.

1. Pick one taskA workflow with a clear start, a clear end and a person who owns the result.
2. Prototype on safe dataA first version using sample or anonymized records, tested against real questions.
3. Test and tuneEdge cases, wrong answers and handoff rules are reviewed in a weekly demo.
4. Launch with limitsPermissions, logging and monitoring are set before real customers use it.

For a quick read on readiness, try our AI readiness checklist.

Cost and what moves it

Projects start at $1,500 and our rate is $30 an hour, both in US dollars. Price depends on the number of integrations, the amount of content the bot must learn, how many edge cases need handling and the review steps you want. Model usage fees are charged by the provider and sit on top. See the AI automation cost guide for typical drivers. GST at 15% may apply to services supplied in New Zealand, so check with your accountant.