AI mockups for builders, landscapers and contractors

The short answer. Take a photo of the customer's garden, room or elevation, upload it to an AI image tool, and describe the finished job in detail: materials, layout, sizes and finish. You get a realistic "after" picture in a couple of minutes. Show two or three options, let the customer pick, then turn the chosen one into a measured drawing and a written quote. The picture sells the job; the drawing and quote are what you build to.

Why mockups win work

Most customers can't read a plan. They can't picture what "Indian sandstone, 600 by 900, laid in a random pattern with a raised sleeper bed" will look like in their own garden. So they hesitate, they get three more quotes, or they pick whoever made them feel most confident.

A picture of their own house, with the new work in it, changes that conversation. It does three things for a trade business:

• It wins the quote. When three firms quote and one shows the finished job, that firm is the one the customer remembers.
• It cuts misunderstandings. Agreeing on a picture up front means fewer "that's not what I imagined" moments halfway through the job.
• It sells upgrades. Showing the same patio with and without lighting, or with porcelain instead of concrete slabs, lets the customer choose the better option for themselves.

Until recently this meant paying a designer or learning 3D software. Today's AI image tools can edit a real photo convincingly, so a mockup takes minutes instead of days.

What you need

A good photo. Daylight, taken from where the customer will usually look at the work: the back door, the kitchen doorway, the end of the drive. Get the whole area in shot and keep the phone level. The better the photo, the better the result.

An AI image tool that edits photos. The general assistants, such as ChatGPT and Google Gemini, can now take an uploaded photo and change parts of it on request. There are also specialist garden, interior and home-improvement design apps built on the same technology. Start with a general assistant you already pay for, and only buy a specialist tool if you find yourself hitting its limits.

Your real materials list. The mockup should show what you can actually supply and build: your supplier's paving range, the cladding you fit, the worktops you buy in. If you describe products you don't stock, you are selling a job you can't deliver.

A five-step workflow that works

1. Photograph the site at the first visit, and take rough measurements while you are there.
2. Write a specific prompt. Describe the change, the materials, the colours, the rough sizes, and what must stay the same. There is an example below.
3. Generate two or three options. For example, a budget version, a mid-range version and a premium version. Keep the camera angle the same so the customer can compare them.
4. Check every image before you send it. Look for things AI gets wrong: steps that go nowhere, fences that change height, drains or manholes that have disappeared, windows that moved. Fix them or regenerate.
5. Send the images with the quote, clearly marked as illustrative, followed by a measured drawing for the option the customer picks.

An example prompt

Vague prompts get vague, glossy results. Specific prompts get pictures that look like your work. For a landscaping job:

"Edit this photo of a rear garden. Keep the house, the back door, the fence and the neighbour's shed exactly as they are. Replace the lawn nearest the house with a patio about 5 metres wide and 4 metres deep, in light grey porcelain slabs, 600 by 600, laid in a grid with narrow grey joints. Add one step down to the lawn in the same porcelain. Add a raised bed along the left fence, built from dark grey oak sleepers about 45 cm high, planted with ornamental grasses and lavender. Natural daylight, overcast UK sky, photorealistic, no people."

The same structure works for an extension (roof style, brick or render colour, window and door positions), a driveway (block paving pattern, edging, drainage channel) or a kitchen (unit colour, worktop, splashback, flooring).

Where AI mockups go wrong

Scale. AI images look right but are not measured. A patio that looks 6 metres deep might be 4 in real life. Always confirm sizes with a drawing before the customer commits.

Things that can't be built. AI doesn't know about drainage falls, damp-proof courses, load-bearing walls or root protection zones. It will draw a level patio flush with the back door or a glass wall with no visible support. You are the expert, so remove anything you wouldn't build.

Planning and building regulations. A picture of a two-storey extension doesn't mean it will get permission. Make it clear the image shows a design idea, not an approved scheme.

Over-promising on finish. AI tends to make everything look like a magazine shoot: perfect lighting, mature planting, spotless paving. Ask for "realistic, newly finished, overcast daylight" and tell the customer what it will look like on handover day versus a year later.

Keep it honest: the advertising rules

Using AI mockups to show customers what their job could look like is fine. Using them in your marketing as if they were real jobs is not. Under the UK advertising rules (the CAP Code, enforced by the Advertising Standards Authority) and consumer protection law, presenting an AI image as finished work you have done is likely to mislead.

• Label mockups clearly, for example "Illustrative design, not a photograph of finished work".
• Keep your portfolio, website gallery and social posts to real photos of real jobs.
• Put a line in your quote or terms saying the images are for illustration, and that the written specification and drawings are what you will build to.
• Ask the customer before you use a photo of their property for anything other than their quote.

Going further: building mockups into how you quote

Once mockups are working for one person in the business, the next step is to make them part of the process instead of a favour someone does when they have time. For a busy trade business that can look like this:

• The website enquiry form asks for two or three photos of the area and a rough budget.
• An automation produces a first set of draft mockups from those photos and your materials list before anyone visits.
• The surveyor checks and adjusts them on site, and they go out with the quote the same day.

That turns a two-week wait for a quote into a two-day one, and customers who already have a picture they like are much more likely to book. This is the kind of workflow we design and build in an automation build. If you're not sure where AI would save your business the most time, start with the free AI scorecard.

Frequently asked

Can AI create a garden or extension design from a photo?

Yes. Current AI image tools can take a photo of a garden, room or house and add a new patio, extension, kitchen or driveway to it, based on a written description. The result is a realistic picture, not a measured design, so it should be followed by a proper drawing before work is agreed.

Which AI tool is best for landscaping or building mockups?

Start with a general assistant that can edit images, such as ChatGPT or Google Gemini, because you probably already have access to one. Specialist garden and interior design apps can be quicker for high volumes. The tool matters less than the photo you start with and how specific your prompt is.

Are AI mockups accurate enough to quote from?

No. They are good for agreeing the look, but they are not to scale and may show details that can't be built. Quote from your own measurements and a written specification, and treat the mockup as a sales and communication tool.

Can I use AI images on my website or social media?

Only if they are clearly labelled as illustrations. Showing an AI image as if it were a job you have completed is likely to breach UK advertising rules. Keep your portfolio to real photos of real work.

Is it safe to upload customer photos to an AI tool?

Ask the customer first, avoid photos that show people, car number plates or the inside of their home unless the job requires it, and use a business account with the tool's training-on-your-data setting switched off where it has one. Treat the photos as you would any other customer information.