By Greg Allan, tradesman and creator of Data First Coaching
Why does a shop drawing revision never seem to reach everyone who needs it?
Because a fit-out runs across three separate places at once, the workshop, the office and the site, and the business is relying on memory and email as shop drawings move between the project manager, the landlord and the crew. Each place can end up working to a different version of the same drawing, and nobody notices until the joinery doesn't fit the tenancy.
A shop drawing on a retail or hospitality fit-out rarely stays still. The landlord's rep wants a service duct moved, the project manager asks for a change to the front counter, the builder flags a clash with another trade on site. Each of those lands as a small revision, and each one needs to reach the workshop before the joinery is cut and the site crew before the install starts. When it doesn't, the shopfitter finds out at handover, standing in front of a defect that was actually a drawing problem the whole time.
The number that tells the truth here is not what anyone remembers emailing. It is the gap between when a drawing is revised and when the workshop and the site are actually working to that revision. That gap is where most shopfitting rework starts, and it only shows up in the job's own data, never in anyone's recollection of who copied who in.
Why doesn't a progress claim match what was actually built on site?
Because variations get instructed verbally on site, built straight away to keep the programme moving, and never formally captured before the progress claim goes out. The claim ends up built from the original quote, not from what the crew actually did, and the builder or project manager pushes back.
After hours work makes this worse. A late call from a project manager to change a shopfront detail before a deadline often gets actioned on the tools, not written down anywhere the office can see. Weeks later, at the next progress claim, the shopfitter is trying to recall what was agreed and what was simply done as a favour to keep the job moving.
The number that actually holds the answer is the total value of variations instructed on a job against what has been formally claimed for it. Most shopfitters are carrying unclaimed variation value without knowing the size of it until handover, by which point the goodwill with the project manager has usually gone cold.
Why are rising costs and AI both landing on shopfitters right now?
Because materials, joinery components and subbie rates are moving at the same time as the tools to run the office are changing fast, and a shopfitter not onto both is absorbing cost increases with the same amount of unpaid admin time as before.
Commercial fit-out work does not leave much room to carry waste. Supplier pricing on laminates, hardware and shopfitting fixtures moves often, and a tight programme with a hard deadline does not forgive a quote built on an old price. At the same time, AI agents are becoming available to do the non-billable work, chasing approvals, tracking variations, matching progress claims to what was built, that used to eat into a shopfitter's evenings. A business not using that is competing against one that is, with the same margin pressure and fewer hours spent fixing it.
What is the difference between AI that answers and AI that works?
Most AI a shopfitter has tried so far is conversational: ask it a question about a shop drawing or a clause in a tenancy agreement, and it answers. That is useful, but it is not what changes a fit-out business. An AI agent is software that does a job the way you would, with your prices, your wording and your rules, and it goes and does the work rather than waiting to be asked.
The difference matters because a shopfitting job does not fail from a lack of answers. It fails from information not moving: a revision not reaching the workshop, a variation not reaching the claim. AI agents built into the business can sit in that gap, carrying a drawing revision through to the people who need it, or holding a variation against the job record until it is costed and claimed, instead of leaving it to whoever remembers.
How can AI agents keep the workshop, drawings and site working from the same version?
They can only do that if the business is working data first, meaning every price, drawing version and variation lives in one central record that the AI agent reads from, rather than scattered across emails, phones and whatever was printed for site that week.
Picture a shopfitter running joinery out of the workshop for a hospitality fit-out. The price of a laminate finish changes with the supplier, and it gets updated in one place, but the AI being used was still reading an old price list as if it were current, so every quote it touched came out wrong. At the same time, variations the foreman logged from his phone on site were never connected to the job's central record, so they simply vanished, never making it to the progress claim.
What changes it is the business keeping all its job data in one place as the single source of truth, with every AI session connected to that central record so the work the crew does from site or the workshop actually gets saved against the right job, not lost in a phone or an inbox.
Should you replace your current software or build something shaped around your own jobs?
Not everything needs replacing. Some off the shelf tools, a supplier ordering system or a basic invoicing package, can genuinely suit a shopfitting business as it stands, and the business's own data is what shows which ones to keep.
The tools most shopfitters feel the pain with are the generic ones built for every trade at once, where a drawing revision, a tenancy clause or a progress claim format has to be forced into a system that was never shaped around how a fit-out actually runs. The alternative is not throwing everything out, it is building systems around your own jobs, your own suppliers and how your retail and hospitality customers actually buy, with AI agents doing the office work inside them rather than renting software built for somebody else's trade.
How does Data First Coaching approach this?
Data First Coaching starts from the shopfitting business's own numbers, run through the Trade Business Intelligence, rather than from the owner explaining the business from memory. For a shopfitter that often means looking first at where drawing revisions and variations actually go missing between workshop, office and site, because that gap is usually the real root of the defects and disputed claims that show up at handover.
Greg Allan, who built his own trade business before coaching others, puts it plainly: “Agentic AI is what we're gonna be talking about here, is you can actually get the AI to go and do work for you.” How that gets built into a shopfitting business is worked through inside the Academy, with members, because the Trade Business Intelligence is powered by every member for every member, and the Academy is built for owners who are open minded, innovative and willing to share what they learn.

