AI design tools showed up in furniture studios about three years ago and have been quietly evolving since. The early generation was novelty. The current generation is something more interesting — a set of tools that actually fit into the design workflow at specific points, and that have started to reshape how designers and clients communicate about a project.

This is a working report from May 2026 on what’s worth using, what’s still not ready, and where the tools are going.

The conversation has shifted

Two years ago the question was “can AI replace the designer.” That question has been quietly resolved in favour of “no, but it changes the work.” The clients who turn up with an AI-generated kitchen layout don’t expect the studio to build it as drawn. They expect it to be a starting point. That’s a healthier conversation than the one we were having when AI image generators first hit the consumer market.

The designers who do well with this shift are the ones who treat the client’s AI sketches as a brief, not a competitor. The designers who struggle are the ones who feel the AI is encroaching on the creative work. The middle ground — using AI tools as a faster way to explore options inside the studio — is where most of the actual productivity is showing up.

Where AI is helping

Concept variation is the strongest use case. A skilled designer can sketch four or five concepts for a custom built-in in a morning. A designer with an AI tool can produce twenty in the same time, all at the level of “rough but readable.” The volume of options widens the conversation with the client. It does not replace the discipline of choosing among them.

Visualisation is the second area. Putting a proposed piece into a photo of the client’s actual room is now reliable enough that most studios doing custom built-ins are using some version of this. The early versions were prone to obvious composition errors. The May 2026 versions are good enough that clients sometimes can’t tell the AI render from a photo until they look closely.

Material exploration is the third. Asking a tool to show the same design in walnut versus oak versus painted poplar, with realistic grain and finish, used to require a real sample and good lighting. Now a designer can put six material variations in front of a client in ten minutes. The client still needs to handle a real sample to commit. But the conversation about which direction to take happens faster.

Specification documentation is the quiet sleeper. AI tools that take a designer’s sketch and produce clean dimensioned drawings, cut lists, and hardware specifications save serious time. They are not perfect. The drawings need a human eye before they go to the workshop. But the time saving is real, particularly on standardised joinery elements.

Where AI is still not helping

Joinery details remain a manual job. The tools can produce something that looks like a dovetail in a rendering. They cannot produce the actual specification of the joinery that the workshop needs. Custom hardware integration, edge details, finish schedules — these are still designer-and-maker work.

Structural decisions are not something I’d hand to AI yet. A client’s request for a bookcase with no visible supports across a four-metre span is a structural engineering question, not a styling question. The AI tools will happily render the bookcase. They will not flag the structural issues that need to be solved.

Workshop planning is also still ahead of the AI tools I’ve tested. Sequencing the build, deciding which pieces get prepped together, working out the finishing order — these are decisions that a workshop foreman makes by walking around the workshop and looking at what’s available. No AI tool I’ve used understands a workshop the way a foreman does.

The integration challenge

The studios that are getting the most out of AI tools are the ones that have thought about integration. The designer’s sketching app talks to the visualiser. The visualiser exports to the documentation tool. The documentation tool produces files the workshop can actually use. Each of those connections is real work.

The studios that have a designer using one AI tool, a draftsperson using another, and a workshop running a third disconnected system are getting some benefit but not the full productivity uplift. The integration is the bottleneck.

For studios thinking about how to approach this, the practical pattern is to pick the workflow stage where the bottleneck is worst, adopt one AI tool that addresses it, and live with that for a few months before adding another. Studios that try to overhaul the whole workflow at once usually don’t finish the project.

What clients are asking for

Clients are increasingly turning up to first meetings with AI-generated concepts they’ve made themselves. The quality of those concepts varies wildly. Some are useful starting points. Others have obvious problems — proportions that won’t work, materials that don’t exist in those configurations, structural ideas that won’t stand up.

The right response from a studio is to take the client’s concept seriously as a brief and then explain what would and wouldn’t work. The wrong response is to either build whatever the AI sketched or to dismiss the AI work entirely. Both extremes are common and both lose business.

Where this goes

By the end of 2026 I expect the tools at the documentation and visualisation end of the workflow will be standard equipment in any custom furniture studio. The tools at the structural and joinery end will still be marginal. The studios that adopt thoughtfully will have a real productivity advantage. The studios that adopt for marketing reasons will have flashy renders and the same workshop bottlenecks they had before.

The piece of design that always was, and will remain, human work is the conversation with the client about what they actually want. The AI tools widen the vocabulary for that conversation. They do not have the conversation themselves.