Short answer
An AI consultant helps an organisation decide where AI is worth using, what it will cost and how to run it safely. Levy Street’s consulting is done by the engineers who also build the systems: we map real workflows with your team, test candidate uses against your own documents, and recommend whether to buy a tool, build something narrow, or leave the work alone.
- Start with one workflow and a number to move, not a company-wide AI strategy.
- Every recommendation comes with a build-or-buy decision and running costs in NZD.
- Advice is grounded in systems we have built and run, from internal agents to products with paying users.
- New Zealand context built in: Privacy Act 2020, data residency and local software such as Xero.
When AI consulting is worth paying for
AI consulting is worth it when you know AI could help but cannot yet answer three questions with confidence: which work to start with, whether to buy or build, and what it will cost to run. If you already know the workflow and the tool, you probably need implementation, not advice.
The organisations we help usually fit one of these situations:
- Pressure without a plan. The board or owners want an AI answer, staff are already using ChatGPT on their own, and nobody has decided what is allowed.
- A stalled pilot. A proof of concept looked good in a demo and then stopped short of production. See AI system rescue if something is already built and struggling.
- Expensive repetition. Skilled people spend hours a week on triage, reporting, document review or research, and the cost is visible.
- A vendor decision. You are comparing Copilot, an industry product and a custom build, and need someone without a licence to sell.
What a Levy Street engagement includes
Every engagement is run by an engineer who builds production systems, alongside the people who do the work. The deliverables are specific:
- Workflow map
- The real process, including handoffs, exceptions, approvals and the systems involved, drawn with the people who run it.
- Opportunity ranking
- Candidate uses scored on value, feasibility, risk and data readiness, with the arithmetic shown.
- Build-or-buy call
- For each candidate, whether an existing product will do, what configuration it needs, or why a custom system is justified.
- Tested prototypes
- Quick experiments on your own documents and data, so the ranking is based on evidence about accuracy, not assumptions.
- Risk and privacy notes
- What personal information is involved, where it is processed, who reviews outputs, and how errors are caught.
- A costed first project
- Scope, success measure, timeline, and running costs in NZD, ready to approve or reject.
How the work runs
- Pick the workflow. One conversation with leadership to choose a workflow with a clear owner and a number you already track, such as turnaround time, hours per week or error rate.
- Sit with the work. We spend time with the people closest to it and collect real examples: emails, forms, reports, edge cases. This is where most of the value is found.
- Test the hard parts. We prototype the riskiest steps first, on your data, and measure how often the output is right. If accuracy is not good enough, you find out now rather than after a build.
- Decide. You get the ranking, the recommendation and the cost, presented to the people who will approve it. Then you choose: buy, build, train, or stop.
What AI consulting costs in New Zealand
The cost of advice is driven by scope: one workflow, a department, or a whole organisation. We do not publish day rates, and we scope every engagement as a fixed piece of work once we understand the workflow. The larger number to understand is what the resulting system costs to build and run, which is usually dominated by integration and review time rather than model fees. Our guide to what custom AI projects cost walks through the arithmetic, and the AI ROI calculator lets you test your own numbers.
Why build experience matters in advice
Advice from people who have never run a system in production tends to underestimate the boring parts: permissions, exception handling, monitoring, and what happens when the model is confidently wrong. Our recommendations come from building and running systems such as:
- A property management agent for APL Property: email triage, records and order processing for a regional property firm, with staff approving what goes out.
- PainPal: an AI workmate for workers’ aches and strains, launched by BetterWorx and CHASNZ, that answers from expert-reviewed guidance and knows when to send someone to a professional.
- Bebop: Blackpearl Group’s AI sales agent for small and medium businesses; Levy Street engineers work with Blackpearl’s AI team on it.
We are also lab agnostic. We use models from OpenAI, Anthropic and Google and open-weight models, and choose by testing against your workflow. See how we choose an AI model.
New Zealand specifics we plan around
- Privacy. The Privacy Act 2020 applies to personal information you put into AI tools, including the rules on use, accuracy and disclosure overseas. Our lesson on protecting customer information when using AI covers the practical steps.
- Data location. Some providers now offer processing in Australian or New Zealand regions; others do not. We record where each system processes data so you can make an informed decision.
- Local software. Many NZ firms run on Xero, MYOB, Microsoft 365, ServiceM8 or industry systems. A recommendation that ignores your existing stack will not survive contact with your team. See our integration guides.
Where we work
We are based in Wellington and work with organisations across the country. Discovery is best done in person, so we travel for it. Local pages: Wellington, Auckland and Christchurch.
In this section
- WellingtonWellington-based AI engineers who map your workflows in person, test on your documents and build what pays off.
- AucklandWellington-based AI engineers who travel to Auckland for discovery, then build remotely with regular on-site visits.
- ChristchurchOn-site discovery in Christchurch from our Wellington team, then remote builds with regular visits.
Written by the Levy Street team in Wellington. Last updated 19 October 2026.