Practical AI consulting for NZ businesses

Most businesses don't need another AI strategy document. They need AI tools that are set up properly, connected to their own data and safe to use. That's the work we do as AI consultants in New Zealand: we choose the models, connect them to your systems, host them where you're comfortable and guide your staff while they get used to them.

We're a software development and DevOps team, so when the AI needs software around it, we build and run that too.

What an AI consultant does for a small or mid-sized business

The useful part of AI work is rarely the model itself. It's the plumbing around it. For a small or mid-sized business, that usually means:

• finding the jobs where AI saves your team real time, and the ones where it doesn't
• choosing the right model for each job, and the right place to run it
• giving AI tools access to the data they need, and nothing more
• bringing AI into your software team for code review, bug fixes and refactoring
• getting staff to the point where they can ask questions of your own data from their AI assistant

How an AI engagement runs

1. We look at how your team works: where the time goes, what data people hunt for, and what would change if they could just ask for it.
2. We propose one small first step, with a clear description of the work, a price and a time frame, before anything starts.
3. We set it up with real staff and real data. If something unexpected comes up, we stop and talk it through with you.
4. We roll out what works, then mentor your team as they start using it day to day.

Starting small means you find out quickly whether AI pays its way in your business, before you commit to anything bigger. When the first step is automating a process, such as sorting email or reading invoices, that's our AI automation service.

Example: bringing AI into a legacy publishing business

We introduced AI into a legacy publishing business, starting with its development team. AI now reviews code, helps get pull requests merged sooner, and assists with bug fixes, refactoring and data analysis.

We then connected the company's data to AI models through an MCP server. MCP (Model Context Protocol) is an open standard for connecting AI assistants to outside data and tools. With it in place, staff can query the company's data directly from their AI agent.

The AI models we work with

We're not tied to one provider. Depending on the job, your budget and where your data is allowed to go, we use:

• OpenAI models
• Anthropic's Claude
• Amazon Bedrock, for models hosted inside AWS
• open models we host ourselves with vLLM, when data can't leave your own network

Mentoring your staff rather than AI training courses

We don't run AI training courses. Courses teach people to use a chat window. We put the tooling in place first, then mentor and guide your staff while they use it on their own work. People learn faster when the AI already knows about their documents, their systems and their data.

AI solutions hosted where you choose

You decide where your AI tools and data live. We can host on-site on your own servers, in a cloud account, or in a data centre of your choice. If a model needs to run without sending anything to an outside provider, we can self-host it.

Privacy questions to ask before your data goes into an AI tool

Ask these before any customer data goes near an AI tool. They're a starting point, not legal advice.

• Where will the data be processed and stored, and for how long?
• Can the provider use your data to train its models? Check the terms for the exact plan or API you're using, not the provider's general marketing.
• Who in your organisation will be able to see what through the AI tool?
• Does the tool handle personal information? If it does, the Privacy Act 2020 applies. The Office of the Privacy Commissioner's guidance on AI recommends doing a privacy impact assessment before you start using an AI tool.

When you don't need an AI consultant

Plenty of businesses don't, at least not yet. If your team mainly wants help drafting emails and documents, a business subscription to an AI assistant may be all you need. If nobody can yet say which job they'd like AI to take on, start by listing the tasks that eat the most time each week. We're happy to tell you on a first call that you're not ready for us.

Where we're worth talking to is when AI needs to reach your own data, your code or your systems, and you want it done properly.

AI experts who also build and host software

Most AI projects turn into software projects quickly. The AI has to connect to your systems, respect permissions, keep running and be looked after. Because we already build custom software and run infrastructure through our DevOps service, the work doesn't stall when the advice runs out.

AI consulting questions we get asked