Choosing an enterprise AI platform in New Zealand.

01What an enterprise AI platform is

An enterprise AI platform is one place to build AI agents, run them against the organisation's data and systems, and govern what they do and spend. It differs from a chatbot licence, which gives people an assistant, and from a model provider, which sells access to models. The platform is where agents are approved, connected, paid for and recorded.

02Where New Zealand stands

Datacom's 2026 State of AI Index, released at the end of August 2026, found 91 percent of New Zealand organisations use some form of AI and 15 percent are scaling it across the organisation. Twenty percent have a dedicated AI budget with clear cost tracking. Among smaller firms, the EMA's survey of more than 300 member businesses, published in July 2026, found 83 percent use AI and 13 percent have an AI policy.

The gap between using AI and running it is what a platform is bought to close. Most organisations are past the first experiment and short of the owner, the budget line and the record that let the second and third deployments go faster.

03The rules a platform has to fit

• No AI-specific Act. New Zealand's AI Strategy, published by MBIE in July 2025, focuses on adoption and applies existing law to AI. MBIE's Responsible AI Guidance for Businesses, from the same month, is voluntary.
• The Privacy Act 2020. The information privacy principles apply to personal information in AI inputs and outputs. The Office of the Privacy Commissioner has published guidance on how they apply to AI.
• Information privacy principle 12. Personal information may be disclosed to an organisation outside New Zealand only if it is subject to the Act, has comparable safeguards, or has agreed to protect it adequately, for example through model contract clauses. Otherwise the person must authorise it after being told the protections may differ.
• Directors' duties. Boards answer for AI under the duties they already hold. Our guide to AI governance in New Zealand covers the questions directors ask.

Sovereignty is rising up buyers' lists. In November 2025, Gartner predicted that by 2028, 65 percent of governments worldwide will introduce some technological sovereignty requirements, and advised CIOs to build data sovereignty into AI strategies from the start.

04Six tests for a New Zealand buyer

1. The data path. Where prompts, files and outputs are processed and stored, whether any provider uses them for its own purposes such as training, and whether the contract gives the protections principle 12 expects.
2. Your models, your cloud. Whether the platform runs in a cloud tenancy you control, uses the models you choose, and lets you change provider without rebuilding your agents.
3. An owner and an approval for every agent. A named person, a recorded scope and a review date, before the agent reaches live data.
4. Cost by team, with caps. Spend attributed to the team and project at the moment it happens, with limits that hold before money is committed.
5. A way out. Export of agents, data and records if you leave, in formats you can use.
6. Someone in your time zone. Support during New Zealand business hours, people who know local rules, and references you can call.

05Questions to ask a vendor

ENTERPRISE AI PLATFORM: QUESTIONS FOR A VENDOR (NZ)

1. Where is our data processed and stored, by region?
2. Does any provider use our prompts or outputs for its
   own purposes, including training?
3. How does the contract meet Privacy Act principle 12?
4. Can it run in our cloud tenancy, with our models?
5. What does it take to change model provider?
6. Who owns each agent, and where is that recorded?
7. How is cost tracked by team, and where are caps enforced?
8. What do we take with us if we leave?
9. Who supports us in New Zealand hours?