AI agency, AI automation agency or AI consultant: which does an NZ business need?

01What is the difference between an AI agency, an AI automation agency and an AI consultant?

An AI marketing agency produces and runs your marketing with AI systems. An AI automation agency builds workflows that take repetitive tasks off your staff. An AI consultant advises, plans and trains while your own people do the work. All three call themselves an AI agency, so the job you need done decides which one fits.

An AI agency is an outside provider that builds, runs or advises on AI systems for a business, usually specialising in marketing output, process automation or strategy and training.

The fourth option sits outside the agency model: a person already on your team runs the work, and an outside engineer builds the first systems and shows them how. It is worth weighing alongside the other three because many NZ owners are already partway there. Xero's May 2026 research found 61% of New Zealand small and medium businesses proactively using AI, and 79% of them learning through self-guided experimentation¹.

If you are still deciding whether to buy outside help at all, the guide Do you need an AI marketing agency in NZ? covers that question. This page assumes you have decided to get help and want to pick the right kind.

02What does an AI marketing agency do?

An AI marketing agency plans, produces and publishes marketing using AI systems it operates, with people reviewing before anything goes out. Typical services are content, SEO and AI search, paid ads, email, social posts and reporting. You buy output and results each month, and the agency's people and systems do most of the work.

The service list overlaps with a traditional agency's. The difference shows in how the work is made: drafts come from systems trained on your business, a person edits and approves them, and the agency can usually produce more for the same fee. Some agencies also work on how your business appears in AI answers, often called answer engine optimisation.

The label covers a wide range, from a traditional agency using ChatGPT internally to a fixed-price system with a dashboard. Ask what the AI does for you and who checks it. The glossary defines an AI marketing agent, and the guide to AI marketing agents sets out what a complete marketing system covers.

03What does an AI automation agency do?

An AI automation agency connects the software you already use and adds AI steps so routine work runs on its own. It sells working workflows: an enquiry answered and routed, a lead scored, a booking made, a quote drafted. It rarely produces marketing content, and it seldom tells you what your strategy should be.

A typical AI automation agency services list for a small business looks like this:

• Website chat or an AI receptionist that answers common questions and books calls
• Lead capture that scores enquiries and sends them to the right person
• Follow-up emails and texts after a quote, a booking or a missed call
• Data moved between your CRM, accounting software and spreadsheets
• Drafts of quotes, proposals or reports from a template and your data
• A dashboard pulling numbers from several tools into one view

Most of this is marketing automation with an AI step added, built on platforms such as Make, Zapier or n8n. The agency's skill lies in choosing the right process to automate and keeping the workflow running when a connected tool changes. The guide to AI marketing automation for small business covers which processes usually pay back first.

04What does an AI consultant or AI transformation consultant do?

An AI consultant reviews how your business works, finds where AI will save time or win work, writes a plan with a sequence and budget, and trains your staff to use the tools safely. An AI transformation consultant does the same across the whole business. Either way, your team does the ongoing work once the engagement ends.

Consulting suits a business that wants to understand the options before spending on builds or retainers, or one where several teams want AI at once and somebody needs to set rules. Good consulting output is concrete: a short written plan, a list of approved tools, a privacy assessment, and one or two working changes your staff already use.

Weak consulting output is a long slide deck. If the plan does not name who owns each step and when it starts, the engagement has produced advice your team cannot act on. The guide to AI marketing transformation shows what a full plan for a small business covers.

05When does an in-house person with an outside engineer make more sense?

An in-house person with an outside engineer makes more sense when someone on your team has a few hours a week and wants to learn. The engineer builds the first systems on your accounts, documents them and trains your person to run them. You keep the knowledge inside the business and pay outside rates only for the build.

The cost logic is simple. A new hire is expensive: SEEK's October 2026 advertised average for a marketing coordinator in New Zealand was NZD 61,498², and Robert Walters' 2026 salary guide puts Auckland marketing coordinators at NZD 60,000 to 85,000³, before employer KiwiSaver, which Inland Revenue confirms rose to 3.5% on 1 April 2026⁴. Using a person you already employ avoids that cost, and the outside build is a one-time spend.

The risk is time. If your person is already at capacity, the systems will sit unused. Before choosing this route, agree how many hours a week they will spend on it and protect those hours.

06How do the four options compare?

The four options differ most in who does the ongoing work and what you own when the engagement ends. Agencies do the work and often keep the process. Consultants leave a plan and trained staff. Automation agencies leave workflows that need upkeep. An in-house person with an engineer leaves both the systems and the skill inside your business.

Plenty of providers mix columns. An agency may offer a consulting phase first, and a consultant may build one workflow as part of the plan. Use the table to ask which column most of your fee pays for.

07How is AI agency work usually priced?

AI agency work is priced in three shapes: a fixed project fee for a build or a plan, a monthly retainer for ongoing work, and an hourly rate for advice. Published NZD prices for AI-specific providers are rare, so the general NZ agency and consultant ranges below are the best public reference available.

What the public sources show, as of October 2026:

• Jelly Digital's September 2026 guide puts NZ agency retainers at NZD 500 to 1,500 a month for one channel, NZD 1,500 to 4,000 for multi-channel and NZD 4,000 to 10,000+ for full service, with hourly rates of NZD 95 to 200⁵. It is an agency's own estimate, not a survey.
• Growin's price page listed a Launch package at NZD 1,500 setup plus NZD 500 a month plus GST, and a Foundation package at NZD 4,000 plus GST one-off⁶, covering ads and lead intake.
• King Tide's July 2026 guide puts NZ consultants at NZD 150 to 250 an hour or NZD 2,000 to 5,000 a month⁷. These are marketing consultants; AI consultants who publish rates are uncommon.

AI automation agencies in NZ rarely publish prices, so ask for a written quote per workflow plus the monthly fee, and check whether software subscriptions are included. Hidden pricing is normal across the region: a June 2026 Digital Nomads HQ audit of 411 service lines across 140 Australian agencies found 78% publish no pricing⁸. The AI marketing cost guide for Australia and New Zealand sets these ranges against hires and tools, and the marketing agency cost guide breaks retainers down by service.

08How do you vet an AI agency or consultant?

Vet any AI provider by asking who reviews the output, what you own at the end, whether the work is documented, how your customer data is handled under the Privacy Act 2020, and what proof you can inspect yourself. A good provider answers all of it in writing within a week, and welcomes the questions.

Questions to ask in the first meeting:

1. Who reviews outputs before they reach a customer? Name the person and the step. This matters because people trust AI output too easily: KPMG and the University of Melbourne found 51% of New Zealand employees had relied on AI output at work without evaluating its accuracy⁹ (surveyed over summer 2024 to 2025). A human in the loop should approve anything published, sent or spent.
2. What do I own at the end? Accounts, workflows, prompts, content and data should sit in your name from day one. Anything that stops when you cancel was rented.
3. Is it documented? Ask for a written description of each system, including what it reads and who to call when it breaks.
4. How is my data handled? Ask which AI tools touch customer information and where they store it. Under the Privacy Act 2020¹⁰, the Privacy Commissioner's information privacy principles include storage and security of information (IPP 5) and disclosure outside New Zealand (IPP 12)¹¹, and the Commissioner expects a privacy impact assessment before AI tools are used, updated regularly¹². If a provider builds enriched lead lists, ask how it meets IPP3A, in force since 1 May 2026, which requires telling people when their information was collected indirectly¹³.
5. What proof can I inspect? A live or recorded run on a real business, with inputs, outputs and the review step visible. Logos and percentages on a sales deck prove little.

Red flags that should slow you down:

• Guaranteed rankings, guaranteed mentions in AI answers or guaranteed revenue
• Accounts set up in the provider's name, or a platform you cannot export from
• No written scope, or a long minimum term before you have seen month-one output
• Results quoted with no client you can contact and no source you can check
• Vague answers about where customer data goes

Value is where relationships tend to break. Setup's 2025 Marketing Relationship Survey found 61% of clients who ended an agency relationship cited dissatisfaction with value¹⁴, from about 100 responses, mostly larger US brands. Writing down what month one will produce protects you from the same outcome.

09What should you check when hiring an AI agency in Auckland?

In Auckland, check the same things you would anywhere, plus two local ones: that the provider is a registered New Zealand business you can meet in person, and that it can tell you which of your data leaves the country. Judge every shortlisted provider on evidence you can inspect, whatever its ranking on a list.

This guide found no published survey that counts or ranks AI agencies in Auckland, so this section gives you checks rather than names. Look the provider up on the New Zealand Companies Office register¹⁵ to confirm the legal entity and its directors. Ask for a meeting at your premises or theirs; a provider building systems for your business should be happy to see how it runs. Ask who will do the work and where they are based, because a subcontracted build changes who checks quality and where your data travels.

For Australian readers the shape is the same, with one legal difference: the OAIC's 2024 guidance recommends that organisations do not enter personal information into publicly available generative AI tools¹⁶, and the OAIC says the Australian Privacy Act generally applies to businesses with annual turnover above A$3 million¹⁷, plus some smaller ones such as health providers. The guide to marketing agency alternatives prices the non-agency routes for both countries.

10Where Flow AI fits

Flow AI is an AI marketing transformation practice: the strategy, the systems and the agents that run a marketing function, engineered end to end. It is the engineered version of the fourth column: AI agents plus one marketing engineer, Amber Lan, who builds the systems on your accounts, documents them and reviews the work before it ships.

The skill library shows the systems openly. It starts with a written scope that names the first system to install. Each build is scoped and priced per project. The sprints page lists what each install covers.

11The short version

Pick an AI marketing agency for marketing run for you, an AI automation agency for one process that wastes hours, and an AI consultant for a plan before you spend. If someone on your team has the time, an in-house person with an outside engineer keeps the systems and the skill in the business.

Whichever you choose, get the price and scope in writing, keep every account in your name, and ask who reviews the work and where your data goes.

To see which option fits your business, talk to Amber.