AI consulting in NZ: what it costs in 2026

In 2026, New Zealand businesses typically pay $150 to $450 an hour plus GST for AI consulting. On those rates, an assessment can cost anything from a few hundred dollars to more than $15,000, and a first project that connects several systems usually costs $10,000 to $50,000, according to one NZ provider's published guide. Businesses with fewer than 50 full-time staff can get up to half of eligible costs back through MBIE's AI Advisory Pilot, with up to $2,500 towards an AI roadmap and up to $12,500 towards implementation support.

How AI consultants charge

Hourly billing suits short advisory work, such as reviewing an AI use policy or giving a second opinion on a vendor proposal. With hourly billing the buyer carries the risk of overruns, so the engagement letter should include a cap. Day rates work the same way for workshops and on-site discovery, and each day should end with a written output you can keep.

Fixed-price assessments are the usual entry point. The consultant spends a set number of days reviewing processes, data and systems, then hands over a prioritised list of use cases with rough costs. Buyers like the format because the fee is agreed before work starts, and the AI Advisory Pilot's roadmap allowance funds this kind of work.

Project pricing covers a defined build, such as an assistant that drafts quotes from CRM data or a workflow that reads supplier invoices into the accounting system. A fixed-price project moves scope risk onto the consultant, who prices that risk into the quote. A fixed quote can therefore come in higher than a time-and-materials estimate for the same work, and the buyer gets a known number in exchange. Retainers pay for ongoing monitoring, model and prompt updates, user support and a set number of improvement hours each month. A retainer makes sense once a system is in production and needs regular attention.

Published rates for New Zealand and Australia

Published figures are thin. No New Zealand government agency or industry body publishes a survey of AI consulting fees, and most of the pages that rank in search are written by consultancies describing the market they sell into. One NZ AI consultancy says in its own guide (updated 9 October 2026) that it hasn't "established a representative, independently verified NZ market rate", and the rest of the evidence bears that out.

The most detailed NZ figures we've seen are in a November 2026 guide from an NZ automation provider. The guide puts independent consultants at $150 to $300 an hour and established consultancies or boutique firms at $250 to $450, with day rates from about $1,200 for an independent to $3,000 or more for a firm, all excluding GST. The same guide prices assessments from a few hundred dollars to more than $15,000, multi-system builds at $10,000 to $50,000, and large projects above $100,000. The author calls these ranges rough guides.

Recruiter data covers individual contractors. Contractor rates set a floor under consultancy pricing, because a firm adds project management, testing and overheads on top. Talent International's March 2026 list of the ten best-paid tech contract roles in New Zealand runs from $1,280 to $1,520 a day, with identity and access management consultants at the top. No AI role makes that top ten.

Australia publishes more. Hays listed AI engineer contract rates of A$900 to A$1,200 a day in April 2026. In October 2026, reporting on the Hays FY26/27 contractor guide put the top of the range at A$1,360 a day for AI engineers, A$1,640 for AI solutions architects and A$2,300 for chief AI officers. In May 2026, ACS Information Age reported an average Australian AI engineer contractor day rate of $1,292, drawn from Hays' Tech Talent Explorer. Before any consultancy margin, the Australian contractor figures cluster between roughly A$900 and A$1,600 a day. Treat every figure as indicative, and compare quotes on what each one delivers.

What drives the price

The state of your data moves a quote more than any other factor. A business with clean, structured records in a modern system can connect an AI tool quickly. A business with years of PDFs, spreadsheets and email attachments has to pay for extraction and cleaning before any model can use the data. The number of integrations comes next, because every system the AI reads from or writes to (accounting, CRM, ERP, job management) adds build and testing time.

Where the AI runs affects both the build cost and the running cost. Calling a public AI service through an API is the cheapest way to start. A private model on your own hardware or private cloud costs more up front, and the extra cost buys control over where personal and commercial data goes. Who uses the tool matters as well. A customer-facing assistant needs testing, guardrails and a human review process that an internal drafting tool can do without. Quote comparisons often leave out change management, which covers staff training, rewritten procedures and checking whether people use the tool. A quote that leaves out training looks cheaper on paper and often gets less use once it's live.

Running costs people forget

Running costs last as long as the system does, so put them in the business case from the start. Licences are the easiest to estimate. Microsoft's NZ pricing page lists the Microsoft 365 Copilot Business add-on at NZ$34.00 per user per month on an annual plan, excluding GST, with a first-year promotional price of NZ$29.14 between 1 July and 31 December 2026. At the standard price, 20 users cost NZ$8,160 a year before GST. The same page says some advanced features have usage-based billing on top of the licence.

Tools built on an AI model's API are billed by how much text they process, so monthly cost rises with transaction volume. Ask the consultant to model the monthly bill at your current volume and at double. Hosting covers the cloud resources or on-premises server that runs the integration, plus storage for documents and logs. Maintenance covers model version changes, integrations that break when a connected system updates, prompt adjustments and security patching. There's no published NZ benchmark for annual AI maintenance as a share of build cost, so ask each supplier for a written estimate. Staff time rarely shows up on a quote, even though someone in the business has to own the system, check its outputs and report faults.

Funding that reduces the cost

MBIE's AI Advisory Pilot is the most relevant programme. The business.govt.nz funding page lists co-funding of up to 50% of eligible costs, with up to $2,500 for an AI roadmap and up to $12,500 for implementation support. To qualify, a business needs an NZBN, fewer than 50 full-time employees, and to be established and ready to explore practical uses of AI. The business pays at least half, applicants compete for limited funding, and the page lists the pilot as open until 30 June 2027 (last reviewed 22 September 2026). To apply, register with the Regional Business Partner Network and talk to a growth advisor.

The pilot launched in January 2026 with $765,000 from existing MBIE appropriations. In May 2026 it expanded from 50 to 150 businesses, and in September 2026 it was extended by six months to mid-2027, with another 120 businesses. Places are limited, so confirm availability with a growth advisor before you count the grant in your budget.

The Regional Business Partner Network also runs a separate Management Capability Development Fund. The fund pays up to 50% of one-to-one training or up to 75% of group training, capped at $5,000 per business each year, for businesses with fewer than 50 FTE. Only approved training from registered providers qualifies. Digital technology and tools is one of the listed training areas, which makes the fund worth asking about for staff AI training. Neither funding page says whether the two programmes can pay for the same activity, so check with your growth advisor.

Privacy and governance shape the scope

The Privacy Act 2020 applies to any AI system that handles personal information. The Office of the Privacy Commissioner's AI guidance, created in September 2023, recommends a Privacy Impact Assessment before you start, updated regularly. The guidance also says prompts and training data must be kept secure, and warns that reusing personal information to train a model may conflict with IPP10. The Commissioner's advice is direct: "If in doubt, we recommend you do not use AI tools to handle personal information." If a project touches customer or staff data, include a PIA in the scope and budget from the start.

The government's Responsible AI Guidance for Businesses was published on 18 July 2025 alongside New Zealand's first AI Strategy. The guidance is voluntary and sets out practical steps for managing AI risk. When a consultant works from the guidance, buyer and consultant share a reference point for governance decisions. If you're looking for providers, AI Forum New Zealand runs an AI Capability directory that maps New Zealand AI companies and individuals. The Forum says inclusion in the map doesn't imply endorsement, so use the directory as a starting list and do your own checks.

How to scope a first project

Choose one process with measurable volume and a clear owner, such as preparing quotes, coding invoices or answering repeat customer questions. Pay for a fixed-price assessment first, through the AI Advisory Pilot roadmap allowance if you qualify. Then commit to a small build with a defined success measure, and expand only after the build has run long enough to produce real numbers. Use this checklist on every quote:

• The process, its monthly volume and its current cost in staff hours are written down
• The quote states the currency, GST treatment and every exclusion
• Deliverables are defined for each payment stage
• Change requests have a written process and rate
• Monthly licence, usage and hosting costs are estimated at current volume and at double
• Maintenance and support are priced separately from the build
• A Privacy Impact Assessment is in scope wherever personal information is involved
• The quote states where data is stored and processed
• Your business owns the code, prompts and configuration when the project ends
• A success measure and a stop point are agreed before the build starts
• AI Advisory Pilot eligibility has been checked with a Regional Business Partner growth advisor

Where DataForge fits

DataForge is an Auckland consultancy based in Hobsonville. We work with New Zealand and Australian organisations on AI consulting, automation, custom software, ERP and IT service management. We build what we recommend, so the same team carries an assessment through to delivery. For businesses that want to keep data in-house, we can set up a private AI server so data stays on the client's own network. We also make our own products, POS Forge and ForgeBox. We quote prices after a free assessment, so the figure you get matches your scope. Our AI consultancy page explains how we work.

Next steps

• Take the free AI Opportunity Score, a 2 minute quiz that shows where AI could help your business
• Book a free 30 minute walkthrough with the DataForge team to talk through a first project
• Register with your Regional Business Partner and ask a growth advisor about the AI Advisory Pilot
• Get written quotes from at least two providers and check each one against the checklist above

More guides

• How to choose an AI consultant in New Zealand
• AI consultants in New Zealand: a 2026 guide
• AI consultants for New Zealand businesses: how DataForge works