Three AI Systems Every Trade Business Needs to Stop Losing Time and Leads

You are up on a roof, or under a sink. The phone rings and you cannot get to it. By the time you call back that afternoon the customer has already booked the plumber who answered first. The quote you sent last Tuesday has gone quiet because nobody checked in on it. And the subcontractor who starts Monday still has not sent through his insurance certificates. None of that is a lack of effort. It is a lack of systems and it is quietly costing you jobs.

In this episode David Jenyns sits down with Zac Nielsen of Elite Systems AI to walk through three trade business automation use cases you could put in place without a technical bone in your body: an AI voice agent that answers every call, a quote follow-up sequence that closes work while you drive and a one-click onboarding pack that ends the document chase. If AI has always sounded useful but impossible to start, this is the conversation that gives you a first move. Worth reading how to systemise your processes alongside it, because everything Zac builds sits on top of a documented process.

In This Episode

• 00:45  Why trades businesses leak time and leads before anyone answers the phone
• 03:45  The $5-an-hour trap versus the work only the owner can do
• 06:00  The dollar per hour ladder, run over your own week
• 08:30  70 calls a day and the ones that never get returned
• 11:30  The invisible cost of a lead that goes cold
• 13:00  Automation, AI automation and AI agents, explained plainly
• 17:15  How one accounting firm cut a 1,000 hour job to under 10 hours
• 21:15  Use case one: an AI voice agent that answers every call
• 29:45  Use case two: quote follow-up that runs without you
• 34:30  Use case three: one-click subcontractor onboarding
• 39:00  The stack behind the businesses that make this stick
• 41:00  Three simple things to set up this week

👤 Today’s Guest, Zac Nielsen

Zac Nielsen is the founder of Elite Systems AI, an automation agency that builds what it calls AI employees: one tool doing one clearly defined job inside a business. His firm works with service businesses up to $30M in annual revenue and a large share of that work is with trades operators, where the wins tend to be blunt and immediate. Notably for a technologist, his first instruction to most new clients is to stop and go document how they already work, because he has watched too many businesses try to automate a process that only ever existed in someone’s head.

He posts a steady stream of practical AI material for business owners on Instagram, where he goes by itszacnielsen.

Website: elitesystems.ai

📋 Trade Business Automation: Where to Start and What to Fix First

Based on the interview with Zac Nielsen, founder of Elite Systems AI.

Zac’s advice on AI is refreshingly unglamorous. Work out what your time is worth. Write down what you already do. Then bring in the technology to run the parts that never change. He puts it in that order on purpose, because the businesses that skip the middle step end up paying for the same work twice.

Work Out What Your Hour Is Actually Worth

Zac opens every client conversation with the same question: how much is your time worth? Not what you invoice, but what an hour of your attention produces for the business. Owners rarely charge themselves an hourly rate, so the answer tends to be a shrug. That shrug is the problem.

“Most founders find themselves focusing on $5 an hour work instead of $500 plus an hour work. And so the whole game of AI automation is that’s exactly what we’re trying to do. We’re trying to free up that time from team members and founders so that they can focus on the highest leverage work in the business.”
Zac Nielsen, 03:45

He draws it as a ladder. At the bottom sits the $5-an-hour work: copying numbers between spreadsheets, chasing a signature, retyping an address. In the middle, around $50 an hour, sits scheduling and quoting. At the top sits strategy, closing deals and the handful of judgement calls that genuinely need the owner. Map your last week onto that ladder and the gap is usually embarrassing.

Dave points out this is not a new idea, just a new tool. Farming got tractors. Manufacturing got assembly lines. Each one was a way of handing the low-value work down a rung so the humans could climb one. AI is the current rung and the same rule applies: you have to be able to describe the work before you can hand it over, which is exactly what makes delegating tasks effectively and automating them the same skill.

Where Trade Businesses Leak Time and Money

Zac describes a client who is on the tools most days, often literally on a roof, taking 70 calls a day and missing about half of them. He cannot answer from up there, so he works through the backlog in the afternoon. By then, ten of those callers found him through a Google search, liked his reviews, got no answer and rang the next name on the list.

The same pattern repeats in the quoting. A quote goes out, nobody follows it up for a week, the customer has since spoken to two other operators and the job is gone. Not lost on price. Lost on silence. Then there is the subcontractor starting Monday who needs insurances, an ABN, a white card, inductions and safety paperwork, which turns into half a day of chasing texts for documents. One of Zac’s clients, doing made-safe work after weather damage, was spending about 50 hours a week on reporting alone. More than a full-time job, on top of the actual job.

Dave’s reading of it is that the worst of these costs are the ones you never see on a statement.

“You might have spent $100 to get your phone to ring on running an AdWords ad, and now you’ve just lost that, and you’re just flushing cash down the toilet.”
David Jenyns, 11:30

If you want to see what the other side of this looks like in a trade, the story of a trades business that built scalable systems is a useful companion to this episode. Same starting point. Very different Monday morning.

Document Before You Automate Anything

This is the part Zac keeps coming back to and it is the reason he and Dave get along. Clients arrive excited about AI and he sends them backwards first.

“They’ll try and automate before they’ve documented it. They try to automate before they’ve actually gotten clear on what all the processes are. And so as a result, they have to back fix everything because it’s just not clear in the first place.”
Zac Nielsen, 20:30

His phrase for the outcome is that you scale the mess. Automation is indifferent to whether the process is any good; it just runs it faster. So an undocumented process becomes a fast, expensive, undocumented process and every fix has to be retro-fitted. Capture what you are currently doing, not what you wish you were doing. The automation step then becomes almost mechanical, because you already know what the system has to do at every point. That is what process documentation is for and it is why AI and automation documentation is the foundation rather than the afterthought.

Automation, AI Automation and AI Agents

Zac splits the technology into three stages, which is genuinely useful if you have been nodding along in conversations without knowing where you sit.

Stage one is plain automation. If this happens, do that. It has been around for decades. Zac notes that when he says the word to business owners, most of them picture a conveyor belt, when what he means is any business process running itself.

Stage two adds intelligence. Quoting is the clean example. Every enquiry is slightly different, so the step needs a bit of judgement before it can produce an output. Drop a language model into the middle of the automation and it can handle that variation.

Stage three is agents. Tools that decide, act and report back. Zac has been building narrow versions of these into businesses for years, but the recent wave of agentic tools pushed the idea into the mainstream and he thinks adoption has hit an inflection point because owners can now see it working rather than being told about it.

His own favourite recent build is a good picture of stage three in practice. A Google Ads agency now gets a structured report generated from how they want accounts managed and it comes through to the account manager as a list of proposed actions in Slack: adjust these keywords, add that negative, disable this bidding rule. The human ticks the boxes like a punch card, approves and the system goes and does it. Zac says it has saved them roughly the cost of a full salary and the client’s reaction was to immediately ask what else could be done the same way.

Dave’s gold standard is Shannon Smit, who runs an accounting firm with a team of around 20 and has been a systemHUB and SYSTEMology client for eight years. Her team used to log into the ATO portal for every client every week, pull the updates, put them into a spreadsheet, review them and send them on. It took roughly 1,000 hours a year.

“They ended up applying some automation and AI and basically took it from 1,000 hours down to less than 10 hours. And it’s just a human in the loop.”
David Jenyns, 17:15

The robot pulls the data, drafts the email and stops. A person reads it before it sends. Shannon has since run the same play across hundreds of tasks, which is where the compounding really happens. The whole story is worth reading in full in her SOP automation case study. She could only do any of it because the process was written down first. She had documented it years before she automated it.

📋 The Three Systems Zac Would Build First

These are the three Zac reaches for with almost every trade business, in this order, because they are the simplest to install and the fastest to pay for themselves. He reckons the three together give most operators back the equivalent of a full-time role and calls that estimate conservative.

1 An AI Voice Agent That Answers Every Call

This is the one that fixes the roof problem. A voice agent picks up every inbound call at any hour. It can also take five of them at once. No receptionist can do that, which is the point Zac makes about capacity: the constraint is not skill, it is how many conversations one human can hold at a time.

What it does on the call depends on the business. It can answer basic questions from a knowledge bank you give it, pre-qualify the caller against criteria you set and either book them straight into your calendar or take a summary and flag the ones that need a human. Zac is blunt that filtering out the tyre kickers is half the value. A trade often needs more triage than a booking, so the realistic version is: qualify, capture and only interrupt you for the calls worth interrupting you for.

He also points out the most common use case is quieter than people expect. Plenty of tradies run the business off their own mobile, so the agent gets wired in as a redirect: it picks up what you cannot, has a real conversation instead of a beep and behaves like a smart voicemail that gets things done.

On speed, Zac cites the figure that 78% of customers buy from the first responder. He adds a wrinkle worth noticing: buyers now research before they call. They have read your Google reviews and looked at your site, so by the time they dial they are close to deciding and the call is a confirmation rather than an enquiry. Miss it and you are not losing a lead, you are losing a booking.

Dave had the experience from the other side while in Florida for a speaking engagement.

“That’s probably one of the first times I’ve had a voice agent where I didn’t touch the human at all. It went all the way through to booking, and then I got to go on the experience without having that.”
David Jenyns, 24:30

On tooling, Zac names Vapi, Bland AI, Atlas and GoHighLevel and says he uses GoHighLevel himself. Where the voice tool sits outside your CRM, something like n8n stitches the two together; some platforms build the inbound call, the outbound call and the agent into one place, which he prefers. If you want a broader tour of where else this fits, the episode on how to use AI in a small business covers the wider ground.

2 Quote Follow-Up That Runs Without You

Zac calls this a no-brainer for every trade business and it is the least technical of the three. A quote gets logged in the CRM and that log starts a clock.

• Day one. A short check-in. Sent the quote through, let me know if you have questions.
• Day three. Not a chase. Something useful, so the touchpoint earns its place.
• Day seven. A gentle nudge that the quote has a use-by date, if yours does.
• Expiry day. A final word before it lapses.

That is the whole thing. It fires on every quote, forever. Nobody has to remember it. Zac’s argument is that close rates lift not because the messages are clever but because they exist at all. Most businesses lose work purely because a busy week swallowed the follow-up. The upgrade, once the plain version is running, is to let AI personalise each message against the original enquiry so the sequence reads like a person wrote it. The same logic drives rebooking past customers on autopilot, which is the version of this pointed at your existing database.

Dave describes where his own team has taken the same idea. Fathom captures the client calls into a repository sorted by client, a style guide and quote format are loaded in, HubSpot supplies the message history and a PowerPoint skill assembles the pitch deck or quote from all of it. Hyper tailored and built out of the same three ingredients: a captured process, a template and a bit of intelligence in the middle.

Both of them land in the same place on how far to take it.

“Start with simple first and then you can always add things later. It’s very easy when you go down this path to start over-engineering things.”
Zac Nielsen, 33:30

Which, as Dave notes, is a core SYSTEMology principle wearing different clothes: capture what you are currently doing, not what you would like to be doing and make that repeatable. That alone delivers most of the result.

3 One-Click Subcontractor and Client Onboarding

A subcontractor starts Monday. He needs insurance certificates, an ABN, a white card, an induction, safety gear and whatever access paperwork the site demands. So somebody spends half a day sending texts, chasing forms and getting things signed. Zac puts it at half a day or more each time. A business bringing someone on every month loses a full day to it and one with a moving roster loses considerably more. Client onboarding is the same job wearing a different hat.

His version is what he calls one-click onboarding and it runs like this. The new subcontractor or client is added to the CRM. That triggers a welcome message with a pack: the forms they need to complete, the documents they need to sign, an onboarding video or briefing note and a clear statement of what happens next. As each item comes back, the system collects it and moves to the next step. A compliance check runs over what has arrived and at its simplest that is AI confirming the fields are filled, the structure is right and nothing is missing. Only then does a human get pinged to review the file and sign the induction off.

Note the shape of that: hands-off until the moment judgement is genuinely required. Same as Shannon’s human in the loop. If you want the customer-facing version of this, the guide on building a video-led client onboarding system covers the pieces worth recording once and reusing forever.

The tooling here is refreshingly dull: GoHighLevel, n8n, Make.com or even Zapier, plus DocuSign or PandaDoc for signatures. Zac’s rule of thumb is to pick the single platform that handles the most of the process and only stitch tools together when one system genuinely cannot cover it.

Dave’s summary of all three is the sharpest line in the episode: automation runs the steps and AI is the glue between them. When a process breaks halfway through, an intelligence layer can work out what you were trying to do and carry it to the next stage, which is exactly the handover that used to need a person.

The Stack Behind the Businesses That Make This Stick

Dave looks at his best-performing clients and finds the same three pieces every time, in order.

Process first. They are clear on how the work really happens before anyone touches a tool.

Then a Systems Champion. Someone inside the business with the visibility and the mandate to capture best practice and build the culture around it. This is the accelerator. What a Systems Champion does for a profitable tradie is the trades-specific version of the role.

Then a provider. A team like Zac’s, or a capability you build internally, to put the technology in.

The order matters because of what happens without the champion. The consultant does excellent work and leaves. The work then sits in a folder on somebody’s desktop until it is forgotten. Dave’s point is that the champion is who makes it get used and that AI helps them enormously because the return is obvious to everyone on the team within about a week. Nobody needs convincing that the admin task they hated is better gone.

Three Things You Can Set Up This Week

Zac keeps the entry point deliberately low, because what he is after is momentum, not a build.

One, start talking to AI. Open Claude, ChatGPT or Gemini and have a conversation about your business. Dave adds that a lot of trades are already paying for Google’s tools and using a fraction of them and that Google Gems will build you a small custom workflow with less fuss than the name suggests.

Two, start recording your meetings. Fathom for calls. Zac now recommends Granola for the in-person ones, which matters for a trade taking notes on site with a prospect. Those recordings are the raw material for every system you document later and a tool like Process Pal turns a transcript into a written system without you starting from a blank page.

Three, connect your website form to your CRM. Zapier’s free tier will do it, in what Zac describes as a two-step job you can look up on YouTube. Add one more step that pings your phone the moment an enquiry lands, so nothing goes into the void while you remember to check later. If you invoice through Xero, the same trick sends a follow-up email the moment a quote goes out.

Zac’s estimate for those three alone is 15 or more hours back. The bigger return is that you find out where the real opportunities are hiding.

The owner who is under a sink missing calls, forgetting to chase quotes and losing half a day every time a subcontractor starts is not short on effort. He is short on systems. What Zac lays out here is that the distance between that Monday and a calmer one is usually three automations and one piece of homework and the homework comes first. Write down how you already do it. Keep the first automation simple enough to be boring. Your business is not broken, your systems are. That is a far better problem to have.

Wondering what “document it first” actually looks like in practice? Here is Dave capturing a business process from scratch, start to finish.

Frequently Asked Questions

Do I need to document my processes before I automate them?
Yes. Zac is emphatic about it. His most common failure pattern is a business automating a process nobody ever wrote down, which he describes as scaling the mess. The fix is dull but it holds: capture what you currently do, then point the technology at it.

Which of the three systems should a trade business build first?
The voice agent, in most cases, because missed calls are the leak with the clearest dollar figure attached. If your phone is already covered, the quote follow-up sequence is the easier build and works on the jobs you have already priced.

Will an AI voice agent annoy my customers?
Both Dave and Zac say a well set up one does the opposite. Dave booked a jet boat ride in Florida entirely through a voice agent and never spoke to a person. Zac adds that plenty of callers now prefer it, because they can get information without committing to a conversation.

What tools does Zac actually use?
For voice: Vapi, Bland AI, Atlas and GoHighLevel, which he uses himself. For connecting things together: n8n, Make.com or Zapier. For signatures and forms: DocuSign or PandaDoc. His preference is always the single platform that covers the most of the process.

How much time does this realistically save?
Zac puts the three systems together at up to a full-time role for many businesses depending on size and calls that estimate conservative. His three starter steps alone he puts at 15 or more hours. The clearest example in the episode is an accounting firm taking a 1,000 hour annual task down to under 10 hours.

Is this about replacing my team?
No. Zac addresses the assumption directly. He frames every one of these systems as moving people up a rung rather than out the door, the same way tractors did in farming. The humans keep the judgement, the nuance and the emotional intelligence. The system takes the repetition.