AI Workflows Agency, case study
AI for NZ Trades and Services: A Workflow Automation Case Study
How a Kiwi electrical and plumbing operator used ChatGPT and Claude to automate quoting and client follow-ups, without changing job management software.
Strategy before tools
The operator we worked with runs a small electrical and plumbing crew across Auckland. Three vans, one office manager, one director on the tools. The pain was not marketing or leads. The pain was the tail end of every job: quotes that took too long to write, follow-ups that never happened, and invoices that came back with client queries because the line items read like site notes.
We did not start with ChatGPT or Claude. We started with a whiteboard. Where was the time actually going? Which tasks were repetitive, text-heavy and low-risk? That short list became the roadmap for the first ninety days.
The three workflows we automated first
Every AI project in a trades or services business should earn its keep on the boring, repeatable work first. Here is where this operator got the fastest wins.
Quoting from a site visit
A voice note from the ute becomes a structured quote in the company template: scope, labour hours, materials, exclusions, and payment terms, ready for a human check before it goes out.
Client follow-ups
Quotes older than three days trigger a warm, on-brand follow-up email. Claude drafts, the office manager approves. Conversion on stale quotes lifted noticeably in the first month.
Job notes into invoices
Rough job notes from the field are cleaned into a professional line-item description customers actually understand, which cuts down on invoice queries and disputes.
Compliance and safety
Toolbox talks, site inductions, and SWMS drafts are generated from a short brief, then reviewed by the licensed tradesperson. Consistency goes up, admin time goes down.
How the quoting workflow actually runs
After a site visit, the director records a two-minute voice note covering scope, access, materials and any risks. That transcript is dropped into a ChatGPT prompt loaded with the company template, tone of voice and standard exclusions. In under a minute, a first-draft quote lands in the office manager's inbox. She reviews, adjusts pricing where needed, and sends it out the same day.
What used to take an evening now takes fifteen minutes. Quotes go out the same day, not the same week, and the close rate on hot leads has visibly improved.
Follow-ups that actually go out
Claude sits behind the follow-up workflow. It reads the original quote, the client name and any notes from the site visit, then drafts a warm, personal follow-up that sounds like the business, not a call centre. Nothing sends automatically. The office manager scans and approves in bulk, which keeps a human in the loop on every client touchpoint.
ChatGPT or Claude for a trades business?
Both, and here is the pattern we see across NZ operators. ChatGPT does the fast, structured work: quote drafts, invoice line items, checklists, voice notes on the road. Claude does the human work: follow-up emails, tricky client messages, safety documents that need a careful hand. Start with ChatGPT on a paid plan. Add Claude the moment you need the tone to feel right.
A first-month plan for a trades or services operator
- Map the top three time-drains in your week. Quoting is almost always one of them.
- Build one company prompt that captures your tone, exclusions and payment terms.
- Run every quote and follow-up through it for two weeks. Refine as you go.
- Save winning prompts in a shared doc so any team member can reuse them.
- Only then look at integrations with Fergus, Tradify, ServiceM8 or your accounting stack.
What changed for the business
Within the first month the director got two evenings a week back. Quotes started going out within twenty-four hours. Stale-quote conversion improved. Invoice queries dropped because the line-item descriptions finally read like something a client could understand. None of that required new software, new hires or a rebuild of their systems. It required a clear view of the bottleneck, and a small set of prompts the team trusted.
Operational Clarifications
Is AI actually useful for a small trades business in New Zealand?
Yes, when it is pointed at the right problems. Trades and services operators usually win the fastest on quoting, follow-ups, job descriptions on invoices, and rostering notes. Those tasks are repetitive, text-heavy, and eat into evenings. That is exactly where ChatGPT and Claude earn their keep.
Do I have to change my job management software to use AI?
No. Start with the web apps and a paid plan. Use AI alongside Fergus, Tradify, ServiceM8, simPRO or whatever you already run. Only look at integrations once you have proven the workflow manually and know exactly what you want it to do.
What about client privacy and compliance?
Use paid tiers such as ChatGPT Team or Claude for Work, where your prompts are not used to train the models. For anything sensitive, strip out names, addresses and job numbers before you paste. Keep a simple written policy for the team so everyone knows the rules.
How long before we see a return?
Most trades and services operators we work with claw back five to ten hours a week within the first month, mostly on quoting, follow-ups and admin. The compounding gain comes when your team stops rewriting the same emails and starts reusing a small library of proven prompts.
Do I need a technical person on the team?
No. What you need is one owner or office manager willing to spend a couple of hours a week refining prompts and sharing what works. A specialist becomes useful only when you want to connect AI to your CRM, accounting or scheduling tools.
Want a workflow map for your trades or services business?
Book an AI Clarity Session. We will map the workflows worth automating first, without the hype, and give you a clear ninety-day plan.
Book a Clarity Session