The 2026 Shaping Business Study found that NZ businesses no longer lack awareness or motivation to adopt AI. What they lack is the capability and confidence to implement it. The question has shifted from "should we use AI?" to "how do we actually make this work?"
This piece answers that question for one specific type of business: a NZ construction or trade business with 10 to 20 people. If that is not you, some of this still applies — but the numbers and workflows are built around that context.
The three tools I am recommending are not new, not complicated, and not expensive. What they are is specific. The time savings below are conservative estimates based on the workflows they replace. Your actual savings will depend on how consistently the tools are used and how well they are configured — which is the hard part, and the part most AI guides skip.
The time cost most construction businesses do not calculate
Before the tools: the why.
A 15-person construction business has roughly four or five people who spend meaningful time on administrative and coordination work — quoting, follow-up, job reporting, client communication, scheduling updates. At an average labour cost of $35 to $50/hour, every hour of that time that can be partially automated or accelerated is worth real money.
The three workflows I see consuming the most unnecessary time in construction businesses at this stage are:
Quoting and proposals. Most estimators spend 60 to 90 minutes per quote writing cover letters, formatting documents, and compiling the same boilerplate they wrote last week. The estimate itself requires skill. The document around it largely does not.
Client follow-up and job updates. Chasing clients for decisions, sending weekly site updates, following up on variations. This is mostly template-based communication done manually because no system exists to automate it.
Internal reporting and job summaries. End-of-week reports, job closeout summaries, meeting notes. Time-consuming to write, low complexity, high repetition.
The three tools below address each of these directly.
ChatGPT Plus at NZ$35/month is the most versatile tool on this list. For a construction business, the primary use cases are quote cover letters and scope summaries, client-facing variation notices, weekly job update emails, and meeting summaries from notes or transcripts.
The key to making this work is not just having the tool — it is having a set of prompts built around your business's language, your standard job types, and your tone. A generic "write me a quote cover letter" prompt produces generic output. A prompt that includes your business name, your typical project type, your client's name, and a three-line brief produces something you can send with minor editing.
The setup investment is about two hours: write five to eight core prompts, save them somewhere the team can access (a shared Google Doc works fine), and train the two or three people who will use it most. After that, it runs itself.
- Estimator completes the pricing. The numbers are done.
- Opens the quote cover letter prompt. Pastes in: client name, project type, key scope items, price, and timeline.
- ChatGPT produces a professional cover letter in 30 seconds.
- Estimator reviews, edits one or two lines, adds to the quote document.
- Total time: 4 minutes instead of 25.
The output is only as good as the input. Generic prompts produce generic results. Spend the two hours building your prompts properly and the tool pays for itself in the first week. Skip that step and you will use it twice and abandon it — which is what most businesses do.
Make.com is an automation platform — it connects your existing tools and makes them talk to each other. At NZ$15/month on the Core plan (USD $9/month, 10,000 operations), it is one of the best value tools available for a business at this size.
For a construction business, the highest-value automation is follow-up. A quote goes out. Three days later, if no response, an email goes to the client automatically. Five days after that, if still no response, a notification goes to the estimator to call. This entire sequence runs without anyone managing it.
The second highest-value automation is job status updates. When a job moves from one stage to the next in your project management tool, Make can automatically send a templated update to the client. No manual email required.
Make connects to over 3,000 apps — if you use Xero, Simpro, Buildxact, HubSpot, Gmail, or Google Sheets, it almost certainly integrates with them directly.
- Quote is sent and logged in your CRM or spreadsheet. This is the trigger.
- Make watches for that trigger and starts a timer.
- After 3 days with no response recorded: sends a templated follow-up email from your address.
- After 5 more days with no response: sends a Slack or email notification to the estimator with the client name and quote number.
- When a response is recorded, the sequence stops automatically.
Zapier is easier to set up and has better documentation. Make is 3 to 5 times cheaper at equivalent volume and handles more complex workflows. For a construction business doing moderate automation — a few sequences, not dozens — Make Core at NZ$15/month is the right call. If you want simpler setup and the extra cost does not matter, Zapier Professional at ~NZ$50/month also works. Either way, one automation tool is enough to start.
The third tool addresses the reporting and documentation problem. For most construction businesses, meeting notes, site briefings, and client conversations are either not recorded at all, or recorded by someone manually typing while trying to listen.
Otter.ai and Fathom both transcribe meetings in real time and produce a summary with action items. Otter is the better general-purpose tool — it works for in-person meetings via a phone app as well as video calls. Fathom is purpose-built for video meetings (Zoom, Google Meet, Teams) and produces better-formatted summaries for that context.
For a construction site manager who does a 20-minute briefing with the crew at the start of each day: open Otter on your phone, hit record. At the end, there is a transcript and a summary. Copy the actions into whatever system you use. That is it.
For a project manager who spends 30 minutes after every client meeting typing up notes: stop doing that. Fathom records the meeting, produces the summary and the action list, and emails it to all attendees. The PM reviews it, edits if needed, and sends in 5 minutes instead of 30.
Otter.ai free plan covers 300 minutes per month — enough for many businesses to start without paying anything. The Pro plan is around NZ$25/month for unlimited minutes.
- Client meeting or site briefing starts. Otter or Fathom is recording.
- Meeting ends. Transcript and summary are generated automatically within 2 minutes.
- PM reviews the AI-generated action list — takes 3 minutes, not 25.
- Actions are added to the job management system. Client receives a summary email.
- No separate "write up the meeting" task exists anymore.
The total cost and what you are actually buying
The $50 to $75/month covers the tool subscriptions. What it does not cover is the setup time — about 6 to 8 hours total to build the prompts, configure the Make automations, and train the two or three people who will use them. That is the real investment, and it is a one-time cost.
The return: 10 to 15 hours saved per week across quoting, follow-up, and documentation. At a conservative $45/hour loaded cost, that is $450 to $675 of freed-up capacity every week. You are paying $75/month to recover $2,000/month in productive capacity. That is the economics of AI done properly.
The one thing that determines whether this works
The tools are not the hard part. The hard part is getting the team to use them consistently.
The businesses that get lasting value from AI tools share one characteristic: they treat the tool as part of the workflow, not an optional add-on. The ChatGPT prompt for quote cover letters is not something people do when they feel like it — it is the way quotes are done. The follow-up automation in Make is not a nice-to-have — it is the standard follow-up process.
This requires two things: documented prompts and workflows that people actually have access to (not buried in someone's notes), and a 30-minute team session to show everyone how it works and why it matters. That is it. No extensive training programme. No change management consultant.
The businesses that fail at AI adoption are not failing because the tools are bad. They are failing because the tools were added on top of existing habits rather than replacing them.
If your quoting process is undocumented and inconsistent — if different estimators do it differently and nobody has agreed on a standard — AI will not fix that. It will just produce inconsistent output faster. The prompt only works if it is built around a defined process. Sort the process first, then automate it. This is why workflow documentation comes before AI implementation in every Blueprint sprint.
Where to start
Do not try to implement all three tools at once. Pick the one that addresses the biggest time drain in your business right now and get it working properly before adding the next one.
For most construction businesses, the order is: ChatGPT prompts first (immediate return, minimal setup), Make automation second (slightly more technical but high-value), Otter last (quick to implement but lower priority than the first two).
Give yourself four weeks per tool. Not because it takes that long to set up — it does not — but because it takes that long to verify that the team is using it consistently and that the output quality is where it needs to be.
After 12 weeks, you have three AI tools running in your business, your team is using them as part of their normal workflow, and you are recovering 10 to 15 hours of productive capacity per week for under $100/month. That is not a transformation. That is a structural improvement — the kind that compounds over time as the tools become habits and the habits become the way the business operates.