AI Solutions for New Zealand Businesses
Built For Habits, Not Hype. Assessment, training, rollout and governance handled together, so AI ends up in daily work rather than demonstrated once and quietly abandoned.
Supporting New Zealand businesses since 1989
What are AI solutions for business?
AI solutions are the tools, training and rules that let a business use artificial intelligence in its actual work: drafting and summarising, answering routine enquiries, handling repetitive steps in a process, and pulling answers out of documents nobody has time to read.
The tools are the easy part. What separates a working AI programme from an expensive one is whether people use it consistently, whether the outputs get checked, and whether anyone decided what data may go into it. AI solutions is the umbrella. Readiness is the preparation, adoption is whether people actually keep using it, enablement is the training and support, and automation is what gets built against a repeatable process. The technology underneath most of it is generative AI: the kind that writes, summarises and answers rather than the kind that predicts a number.
Built For Habits, Not Hype
Most AI projects fail the same way. The tool gets deployed, there is a launch session, everyone is impressed, and within a few weeks usage has quietly returned to nearly nothing.
Nobody decides to stop. It is just that the old way still works, the new way takes a moment of thought, and under pressure people revert. Meanwhile the licence keeps billing, results stay inconsistent, and risk goes up rather than down, because the few people still using it are doing so without much guidance.
A tool that gets opened once is not an AI programme. The thing you are actually buying is a change in how people work.
The honest version
When AI is not the answer
The readiness assessment is allowed to conclude that you should not do this yet. These are the situations where that is usually the finding.
- The process itself is broken. Putting AI over a process nobody has fixed makes the same problem arrive faster and with less visibility. Fix the process first, then decide whether it still needs AI.
- The underlying data cannot be trusted. An assistant answering from records that are out of date or incomplete can give a confident answer built on bad information. Repair the source first.
- The work happens rarely. Automating something that runs a handful of times a year usually costs more to build and maintain than it saves.
- A predictable rule would do the job. If the task follows the same rule every time, a rule is cheaper, faster and easier to audit than an AI model.
- Nobody owns the output. If no one is accountable for checking a decision, that decision should not be automated, however capable the tool is.
- The risk outweighs the benefit. Where a wrong answer carries real cost, keep a person in control and use AI for the drafting rather than the deciding.
We would rather tell you one of these than sell you a programme that quietly stops getting used.
Choosing the tool
Does this need AI, or just automation?
Not every automation needs AI. The difference changes the cost, the risk and how much human oversight the process needs.
Rules-based automation
Rules-based automation follows defined steps and produces a predictable result from a known input: approving a request, sending a notification, moving a record between systems or generating a document from a template.
It is easier to test, easier to audit and usually does not need somebody reviewing every individual result. If the task follows the same rule every time, this is usually the steadier answer.
AI-assisted automation
AI earns its place where the work involves language, interpretation, unstructured information or variation that fixed rules cannot cover cleanly: summarising a long thread, drafting a reply, finding an answer across documents that were never structured for search, or handling enquiries phrased differently every time.
The trade-off is variability. AI-generated output may need human review, particularly where an incorrect answer could affect a customer, a financial decision or another important business process.
Where a predictable rule will do the job, we will recommend the predictable rule. See AI workflow automation.
What we do
What AI solutions include
Six parts, delivered as one programme, each with its own page for the detail. In partnership with Avocado AI, a New Zealand specialist in AI training and workplace enablement.
AI readiness assessment
A look at your systems, data quality, access controls and where the team actually is. You get prioritised use cases, a view of the risks, and a staged plan rather than a guess about where to begin.
Training and enablement
Role-based sessions built around your tools and your work, not generic AI theory. The aim is repeatable habits and the judgement to check an output before acting on it.
Tool rollout and adoption
The right tools for your workflows rather than whatever is trending. Microsoft 365 Copilot is a common fit for businesses already on M365, configured properly with access boundaries and staff onboarding.
Workflow automation
Approvals, onboarding, service requests, reporting and document handling. Every automation starts by mapping the current process, because automating a messy one just makes the mess faster.
AI agents and assistants
Tools that answer routine enquiries, prepare work, gather information, or take controlled actions inside a process you have agreed. Useful where the volume is high and the answers are knowable, with a person still accountable wherever judgement matters.
Governance and safe use
Access boundaries, an approved tool list, data handling rules and written guidance on what staff may and may not put into a prompt. The part that is skipped most often and matters most.
The process
How an AI programme runs
Find out whether you are ready
Systems, data, access controls and the people. Businesses with untidy permissions or scattered documents get very little out of AI until that is dealt with, and it is better to know that before buying licences. This runs alongside our wider IT assessment where that is useful.
Agree the rules before the tools
What is approved, what data may go in, who may use what, and how outputs get checked. Written down and short enough that people read it. See AI acceptable use policy.
Train first, or alongside
Not afterwards. Training that lands before or with the rollout is one of the biggest differences between a tool people use and a licence people forget. This is where the Avocado AI partnership does the work.
Pilot with one team and one job
A narrow use case with a real result, rather than switching it on for everybody at once. If it does not work for one willing team, it will not work for five reluctant ones.
Then keep at it
Usage gets looked at, the people who have stopped get asked why, and the guidance gets updated as the tools change. Adoption fades without this, which is the whole reason it is part of the service.
Cost
What do AI solutions cost in New Zealand?
The readiness assessment is a fixed fee, quoted before it starts. Everything after it is scoped to what the assessment found, because a business with tidy data and a willing team is a different job from one without.
What moves the number:
- How many people need training, and how many distinct roles they cover, since role-based sessions are not one generic session repeated
- What state your data and permissions are in, because a tool like Microsoft 365 Copilot can surface anything the user already has permission to open, and untidy access is a project of its own first
- Whether you already hold the licences, since some of what businesses want is already included in what they are paying Microsoft
- How much you want automated, which is the part that scales with process complexity rather than headcount
- Whether governance exists yet or has to be written from nothing
Worth saying plainly: the licences are rarely the expensive part. The cost that decides whether this works is the training and the follow-up, and that is exactly the line most quotes leave out. It is also where the return on investment comes from, because the ROI on a licence nobody opens is precisely nothing.
Adoption is the product, not the extra
- Adoption is designed in. Training and reinforcement are part of the programme from the start rather than a workshop bolted on after go-live.
- Safer, governed AI. Access boundaries, approved tools and data handling rules go in before people start pasting things into prompts.
- One partner end to end. The same team that runs your managed IT does the AI work, so access, licensing and security are not three separate conversations.
- Vendor-agnostic. We recommend what fits your workflows, data boundaries and budget. Sometimes that is Copilot. Sometimes it is nothing yet.
- Specialist training partner. Avocado AI are New Zealand specialists in AI training and workplace enablement, so the teaching is not done by whoever was free that day.
- Since 1989. We have watched a lot of technology arrive claiming to change everything. That is useful judgement to have on your side of the table.
Go deeper
The detail behind each of these
Everything above is covered at length elsewhere on the site. Start wherever your question sits.
Getting people using it
The tools themselves
Rules and risk
Where we work
AI solutions in Christchurch and Dunedin
Training in the room
Teams in Christchurch since 2009 and Dunedin since 1989. Training works better in person for the first sessions, and both offices can do that. See IT support Christchurch and IT support Dunedin.
Nationwide remotely
For businesses elsewhere in New Zealand the assessment, rollout and follow-up all run remotely, with sessions delivered online. On-site visits where a pilot or a workshop warrants one.
Questions
AI solutions FAQs
What do AI solutions cost in New Zealand?
The readiness assessment is a fixed fee, quoted before it starts. Everything after it is scoped to what the assessment found. What moves the number is how many people and roles need training, what state your data and permissions are in, whether you already hold the licences, how much you want automated, and whether any governance exists yet. The licences are rarely the expensive part, and the return on investment comes from the training and follow-up rather than the software.
We tried AI tools before and nobody used them. Why would this be different?
That is the most common situation we are called into. Tools fail when they are deployed without training, without being fitted to an actual workflow, and without anyone checking afterwards whether use continued. Training comes before or alongside deployment here rather than after it, and usage gets reviewed so adoption does not quietly fade a month in. See AI adoption.
Are you vendor-agnostic, or do you only do Microsoft Copilot?
Vendor-agnostic. We recommend what fits your workflows, data boundaries and budget. Microsoft Copilot is a common answer for businesses already on Microsoft 365 and we enable it properly when it is the right fit, but it is not the only option and sometimes the honest recommendation is to wait. See Microsoft Copilot.
What happens to our business data when staff use AI tools?
It depends on the product, the licence and how it is configured, which is why the approved tool list matters. There are four separate questions worth asking of any tool: what it can reach, where the information is processed, how long it is kept, and whether personal or confidential information should be going into it at all. Microsoft 365 Copilot does not use your prompts, responses or Microsoft Graph data to train its foundation models, and the business and enterprise tiers of the major AI platforms generally exclude workspace content from training by default, but settings and terms differ between products and between versions of the same product. We set access boundaries, define what may and may not go into a prompt, and put that in writing so staff are not deciding it individually. Under the Privacy Act 2020 you remain responsible for personal information your staff put into any tool, which makes the approved list a compliance matter rather than an IT preference. See Privacy Act and security.
What is the Avocado AI partnership?
Avocado AI is a New Zealand specialist in AI training and workplace enablement. The partnership means the training is delivered by people who do that for a living, combined with our side of it, which is the implementation, access control and security. Adoption-first only works if both halves are done properly.
Do we need an AI policy before we start?
Before you roll anything out widely, yes. Staff are almost certainly using AI already, with or without approval, so the practical question is whether there are rules covering it. A short, readable policy naming approved tools and what may be shared is more useful than a long one nobody opens. See AI acceptable use policy.
Is our business too small for AI solutions?
Size matters less than repetition. Plenty of small businesses have a clearer case than larger ones: a ten-person firm answering the same twenty enquiries every week will get more out of this than a hundred-person one whose work is all bespoke. The readiness assessment exists partly to tell you when the answer is that there is not enough repetitive work to justify it yet.
How do you get started with AI solutions?
With the readiness assessment. It looks at your systems, data and access controls, identifies where AI would actually pay off in your business, and gives you a staged plan covering rules, training, a pilot and rollout. Fixed fee, and useful on its own even if you go no further.
Not sure whether AI is worth it for your business?
That is a reasonable place to start from. Tell us what your team spends time on and we will tell you where AI would help, where it would not, and whether you are ready for it yet.
Christchurch 03 343 3124 · Dunedin 03 479 2941 · Nationwide 0800 396 337
About us
AI and IT from one team in Christchurch & Dunedin
Originally formed in 1989 as Willis White & Co, Exodesk was established in Dunedin by Chris Willis and Andrew White. We expanded to Christchurch in 2009, growing into one of New Zealand’s most established IT providers. Today we deliver managed IT services, cyber security, cloud solutions and business phone systems to businesses across Christchurch, Dunedin and beyond.

Contact us
For an AI readiness assessment or a conversation about where to start
Prefer to talk?
Christchurch
Level 1, 85 Riccarton Road
Christchurch 8011
03 343 3124
Dunedin
Level 3, Bartons Building
2 Stafford Street
Dunedin 9016
03 479 2941