A building company owner should start with AI automation by running a two-week time audit, sorting the recurring tasks they find into Delete, Automate, or Delegate, and routing the Automate column through an AI tool. In every workflow, the AI tool prepares the work and flags the gaps. The owner reviews and decides what gets actioned.
A few years back, one of our Executive Business Coaches sat down with Scott Groschel, who runs SG Construction in California, to run a time audit.
Scott had a feeling he was the bottleneck in his own business...
The audit confirmed it.
What Scott found over those two weeks surprised him. He didn't need to do daily site visits to every project, he didn't need to meet with every customer every day, and there were whole categories of work he'd been doing out of habit that didn't actually require him at all.
To start, Scott hired an office manager in the Philippines. Within twelve months, he'd grown the business enough to add three new local hires off the back of that single offshore role.
Today, a lot of the work Scott offshored back then is within AI's capability... pulling data into a CRM, drafting invoices, running pre-meeting briefs, auditing invoices against scope.
Which raises the question we hear from a lot of building company owners right now.
Where do I actually start with AI in my business?
The answer is the same process Scott followed, with the lever at the end shifted from offshoring to AI. There's just one important caveat you need to bear in mind before you commit to any of it.
Because most building company owners don't know where their week actually goes.
They have a sense of it, but the specifics are usually wrong by a wide margin. A time audit replaces that rough sense with a real record.
The mechanics are simple. For two weeks, you write down what you're doing every time the task changes. That might look like five minutes reviewing a quote, twenty minutes on a phone call with a sub, then forty-five minutes responding to client emails.
By the end of week two, you'll have a picture of where the hours go.
What you'll typically find is that 30 to 40 percent of your working week is spent on tasks that either shouldn't be happening at all, or shouldn't be your job specifically. Based on APB's experience coaching thousands of residential building company owners, that range holds up across markets and revenue brackets. [external link: APB SORCI Report]
The owners who get systematic about removing that 30 to 40 percent are the ones who can actually grow.
A time audit also gives you a specific question to answer with AI instead of an abstract one. The question becomes "what task is taking the most of my week that AI can now handle?", and that's a question you can act on directly.
The DAD framework, short for Delete, Automate, Delegate, sorts the tasks the audit turns up into three categories of action.
Delete is the easiest category and the most overlooked. These are tasks that don't need to be done at all, including the daily site visit that's really a habit, the weekly report no one reads, and the customer touchpoint that adds nothing to the project.
Automate is where AI is now doing most of the work in a building company. These are tasks that follow a repeatable pattern, take recognisable inputs, and produce a recognisable output. The kinds of work that used to require Zapier setups or CRM rules are now within AI's capability, with judgement applied at each step.
Delegate is the category that's changed the most for builders thinking about AI. In a building company today, delegation can mean hiring a person, or it can mean setting up a workflow that runs without one, with the work still getting done and no payroll cost attached if the workflow handles it.
Yes, with one important nuance.
Small admin automations like a morning brief or a quote review can deliver value in any building company, including one that's still working through its systemisation. Larger workflows, the ones that handle a full process from end to end, only work properly when the underlying system is already documented and consistent.
AI amplifies whatever runs underneath it.
If the estimating process is chaotic, automating it produces faster chaos rather than fixing the problem. Systemising the process first means the automation produces something everyone on the team can audit, follow, and improve over time.
The practical implication is sequencing. Start with the time audit and the DAD framework, automate the discrete admin tasks where the inputs and outputs are clear, and use the time you free up to document the systems that aren't quite there yet.
Before going large with AI, get the underlying systems written down properly. APB's free Systems Blueprint for Builders is a starting point for that work, and it's the foundation AI sits on top of in any building company looking to grow.
The highest-volume admin work in a building company falls into a handful of categories that map cleanly onto AI workflows. This list isn't exhaustive, but it covers most of what eats a building company owner's week.
Reviewing quotes before they go to clients. AI can read the draft quote, compare it against the scope brief, and flag missing line items, unclear allowances, scope exclusions, and items that look underquoted. The owner reviews each flagged item and decides what to address before the quote goes out.
Drafting variation orders. AI pulls the change details from the source, including site instructions, client emails, and supplier quotes, and produces a draft variation order. The owner reviews the draft, checks the figures, confirms the scope interpretation, and approves it before it goes to the client.
Auditing subcontractor invoices. AI extracts the line items from an invoice, compares them against the approved scope, and produces a variance report flagging discrepancies, potential overcharges, out-of-scope items, and unapproved extras. The owner reviews the flagged items and decides which to query.
Compiling a morning brief. AI pulls the latest activity from your CRM, calendar, recent emails, and any project tracking system in use, and produces a one-page summary of where every active project stands. The owner reviews the brief at the start of the day and decides what needs attention first.
Building reusable workflows. Once a manual process has been run a few times, AI can be set up to repeat it consistently. The owner defines the steps and the checkpoints, AI executes, and the work runs the same way every time.
In every case the framing stays consistent. AI prepares the work and flags the gaps, and the owner is still the final node in any decision that affects scope, cost, or client communication.
The morning brief, for most builders.
It's the workflow with the lowest setup cost, the highest daily payoff, and the cleanest visibility into whether AI is actually saving time.
Most building company owners start their day in three or four tools at once. The CRM has the active job activity, email has the client and sub messages, the calendar has the meetings and site visits, and the accounting platform has the invoices and payments. Pulling a coherent picture of "where are we right now" from those four places takes 30 to 45 minutes and uses up the part of the morning when the brain is freshest.
AI can pull from all four sources, produce a one-page brief, and have it sitting in your inbox by the time you sit down. The setup is a single conversation that takes less than an hour, and the workflow runs every working day from that point forward.
The morning brief also gives you a daily signal that AI is doing what it said it would do. That trust matters once you start considering the harder workflows like quote reviews, invoice audits, and variation orders, which are easier to commit to once you've watched a simpler workflow run reliably for a couple of weeks.
Every AI workflow needs human review, and that's the step that makes the workflow safe to use in a building company.
The pattern is consistent in every workflow. AI prepares the work, flags the gaps, and drafts the output. You then review what's been prepared, make the call, and either approve it or send it back.
The principle stays the same in each use case. Final decisions on cost, scope, and client communication sit with the owner, and AI prepares the inputs.
The path is sequential, starting with a two-week time audit and then sorting what you find into Delete, Automate, and Delegate. From there, pick the one workflow that's eating the largest share of your week, set up AI to handle it, and watch how it runs for a couple of weeks before moving on to the next.
If you've never run a time audit, that's the place to start.
Spend the next two weeks tracking your time, then look at what comes back. You'll find the answer to "where should I start with AI" sitting in front of you on the spreadsheet.