By Greg Allan, tradesman and creator of Data First Coaching
Why does most AI advice for builders feel useless on site?
Most of what gets written about AI for building businesses describes a chatbot: type a question, get an answer. That is not much use to a builder trying to keep slab, frame stage and lock-up dates lined up with a subcontractor who has not confirmed, or a certifier sign off that is holding up the next progress claim. The problem on a building job is never a lack of information somewhere online. It is admin that has to be done, not a question that has to be looked up.
A builder juggling several jobs, each with its own contract, its own variations and its own prime cost allowances, does not need a search engine bolted onto the business. The quotes, the material pricing, the site supervisor's notes and the subcontractor schedule are already sitting in the builder's own systems. The gap is someone, or something, to act on that information the way the builder would, not to answer a question about it after the fact.
Where does a builder's time actually go between the quote and the handover?
A large share of a builder's week goes on work that never appears on a progress claim: chasing a subbie for a start date, re-checking a prime cost item against a supplier price list, following up a certifier, or rewriting a variation because the first version did not match what was agreed on site. None of that is billable, and all of it has to happen for the job to move from quote to handover without a defect or a dispute at the end.
This is where the real cost sits, not in one bad quote but in the small gaps between fixed price and cost plus jobs, between a renovation and a new home build, where something gets missed because nobody was briefed to catch it. A builder working from memory, or from a spreadsheet that is a week out of date, is making scheduling and pricing decisions on an old picture of the job.
What is an AI agent, and how is it different to a chatbot?
An AI agent is software that does a job the way you would do it yourself, using your prices, your wording and your rules, rather than software you have to ask a question and then act on the answer yourself. That difference matters for a builder because the work that eats your week is not research, it is doing: updating a variation, flagging a prime cost change, nudging a subbie before a date slips.
Think about how you would brief a new site supervisor or a new estimator joining your building business. You would walk them through how you quote, how you handle a variation, what a progress claim should look like before it goes out, and what happens when a certifier flags something at frame stage. An AI agent needs the same briefing. Set up well, it behaves like that team member: it knows your rules and gets on with the work, rather than waiting to be asked.
How can AI agents handle the pricing side of fixed price and cost plus jobs?
AI agents can be set up to build out your pricing from supplier catalogue data, which matters for a builder juggling prime cost items and material pricing across both fixed price and cost plus contracts. Instead of an estimator manually checking supplier price changes against every job in progress, an agent can keep pricing current and flag where a prime cost allowance is drifting away from what was quoted.
That does not replace the estimator's judgement on a complex extension or a tricky renovation. It takes the repetitive checking off their plate, so the time that used to go on re-pricing goes back into quoting new work and managing the jobs already on site.
What could this look like in a building business?
Picture a builder running a steady mix of extensions, renovations and new homes, growing fast enough that the admin has started eating evenings that should be family time. Social media is the kind of job that always gets pushed to the bottom of the list, even though it brings in enquiries, because there is never a spare hour between site visits and progress claims to write a post.
That builder sets up an AI agent to write the business's social media posts in the builder's own voice, using real jobs and photos from site. The posts still get checked and scheduled by hand for now, but the next step is having them go out on their own, so that non-billable work stops competing with the jobs that actually need the builder's attention.
What does “data first” mean for a builder's AI setup?
Data first means the builder's own numbers decide what an agent should be set up to do, not a guess about what might help. A building business carries a mix of fixed price and cost plus jobs, each with its own margin, its own variations and its own subcontractor costs, and that picture changes every week as jobs move through slab, frame stage, lock-up and handover.
Looking at that data first often shows the real bottleneck is not the one everyone assumes. A builder might think the problem is slow progress claims, when the data shows it is actually unconfirmed subcontractor dates causing the delay further back. Put an AI agent in front of the right point in that process, and the job cannot move on until the information it needs is actually there.
How does Data First Coaching approach this?
Data First Coaching starts from a builder's own job and finance data through the Trade Business Intelligence, looking at where fixed price and cost plus jobs actually lose time and margin, before any talk of AI agents or systems. Greg Allan built his own building business before building this method, and treats AI the same way he would treat any new person on site. As Greg puts it, “Think of AI as well as I see these agents as another staff member”, which only performs well once it is properly briefed on how the business runs. The how, the actual setup, is worked through inside the Academy, with members, because the Trade Business Intelligence is powered by every member who joins, for every member who joins.
Looking for a business coach who starts from your numbers? Builder Business Coach or every answer for builders.
