AI Consulting for Small Businesses

Most small business owners we talk to aren’t asking whether AI matters. They’re asking a harder question: where would it actually help my business, and what would it cost to find out? That’s the job of a small business AI consultant, and it’s worth being concrete about what the work looks like, because the gap between AI marketing and AI results is wide.

This article walks through what an AI consultant does, what a real engagement looks like from first conversation to working tool, and what New Zealand businesses have got out of it. The examples are our own clients, named, with numbers.

What does a small business AI consultant actually do?

The short version: find the handful of places in your business where AI genuinely pays for itself, then build the smallest thing that proves it.

In practice that means sitting with your team and mapping where time goes. Repetitive data entry, answering the same internal questions over and over, chasing information that lives in one person’s head, reconciling documents by hand. AI is very good at a narrow set of jobs, and a consultant’s value is knowing which of your problems fall inside that set and which don’t.

It also means being told “no” sometimes. Plenty of problems that look like AI problems are process problems, and building a clever tool on top of a broken process just automates the mess. A consultant who won’t tell you that is selling software, not advice.

What does a real engagement look like?

Here’s one from our own casebook. Waterware, a family-owned plumbing and heating importer with a team of about 35, had decades of technical product knowledge concentrated in a few key people, particularly their director Darren Yley. If someone was on leave, colleagues couldn’t get answers. The uncomfortable question was what would happen if a key person left for good.

We built them Gandalf, an internal AI chatbot trained on their own knowledge base. Staff with limited product expertise can now ask a question in plain English and get a clear answer: what the problem is, what the fix is, the spare part number, whether it’s in stock, and the shipment tracking if it’s already on its way. One interface, no expert required.

The part most owners find surprising: from concept to first live working model took four to five weeks. Darren’s words, not ours: “I found the process very easy. From concept to first live working model it was like four or five weeks. I think it was amazing.”

The same client later pointed us at their finance team’s sales order and PO reconciliation process, which was manual, slow, and error-prone. We built an automation that reads supplier PDFs, raises the sales orders, and reconciles purchase orders against invoices. Less re-keying, fewer entry errors, and the finance team’s time back for work that needs judgement.

How much does AI consulting cost?

Less than most owners expect, because the right first step is small. We deliberately start with a scoped engagement — our AI workshop runs half a day and produces a prioritised list of automation opportunities specific to your business, so you know what’s worth building before you spend anything on building it.

The Waterware chatbot is a useful benchmark: a few weeks of focused work, not a six-month transformation programme. If a consultant’s first proposal is a large upfront commitment before anything works, be sceptical.

What should you look for in an AI consultant?

A few things separate useful consultants from expensive ones:

1. They know your infrastructure, not just the AI. Tools have to connect to your actual systems — Microsoft 365, your job management software, your accounting platform. We come at AI from twenty-plus years of managed IT, which means the security, access, and data questions get answered properly instead of discovered later.
2. They start with a small, real deliverable. A working tool in weeks beats a strategy document in months.
3. They can name their clients. Ask for examples with the client’s name attached and outcomes you can check. Ours are on our case studies page — Waterware is happy to be asked about theirs.
4. They think about the people, not just the tool. Gandalf works because the whole team can use it, not just the technically confident few. Adoption is the difference between a tool and a demo.

Is your business ready for this?

If your team spends hours a week on tasks that follow the same pattern every time, or critical knowledge lives in one or two heads, you have candidates. You don’t need clean data, a big budget, or an IT department to start — you need one well-chosen problem.

Talk to us about where AI would genuinely help your business, or come along to an AI workshop and leave with a prioritised list you can act on either way.