Most in the Interiors world are still just prompting. You type an instruction, the model returns an image or a paragraph, and you accept it or you don’t. Visuals get generated, proposals drafted, a sofa sourced from a wider catalogue than anyone could realistically search by hand. It’s faster than it was, but it’s also a fairly transactional way of using something that could be doing far more. I think a lot of the industry is, to some degree, treating a potential creative collaborator like a vending machine.
The rest of the economy has already moved past that point. Advertising, consulting, law and engineering are reorganising around what the recent reports keep calling “agentic” AI systems that hold context, take a goal, work across several steps, and come back with something. The professional’s job in those industries is shifting towards curating and orchestrating rather than producing the work itself.
Clients arrive armed
The first place I notice the shift is the meeting room. Clients used to walk in with Pinterest boards. They now walk in with full research dossiers built through ChatGPT, Claude or Gemini, mood boards generated overnight, and surprisingly forensic critiques of last week’s revision. They cross-check our specifications in real time against what a model told them at midnight. The AI report a client hand-over is the start of the work, not the end of it. The designer’s first job has shifted. It used to be gathering information. Now it’s reading what the client has already gathered, catching what the model has confidently got wrong, and pulling the conversation back to the things only a person who’s stood in the room can see.
The training data problem
The sameness on Instagram isn’t just because every studio is reaching into the same handful of models. It’s because those models were trained on a narrow slice of design , Pinterest, Instagram, listing shots, magazine spreads. That isn’t design data but photography data, optimised for a thumb-stopping moment in a feed. What the model has never seen is how people actually live in a room. How a family moves through a kitchen at seven in the morning. Which corners get used and which never do. That’s the real gap, and better prompting doesn’t fix it.
You can already see this in the market. Proposals, mood boards, finished interiors, all starting to look samey-samey. The pushback is starting too, high-end residential is leaning back into visible craft, makers’ marks, materials with real provenance. Where clients are paying for genuine difference, creative thinking is how you stand out; the tools are just for speed. Visible craft, makers’ marks, irregular handwork are coming back more into demand.
The unglamorous superpower
The unromantic use of AI is the one actually changing the economics of a small studio. Pulling a finish from a workshop in Kyoto, cross-checking a brief against ESG, finding cheaper alternatives, it all happens at speed now, and the reach is effectively unlimited. On top of that, being able to read through the genuinely enormous volume of documentation and regulation a project generates these days, and to draft the planning documents, specifications and schedules of work that used to take days. A four-person studio operating credibly against a global supply chain, and against the regulatory weight that’s grown around the work, in a way it really couldn’t a decade ago.
The people problem
Junior designers used to learn by doing the unglamorous work. Drawing, redrawing, pulling samples, reading a hundred specifications until somewhere around the hundredth, the first one finally made sense. That’s how taste and creativity and design is learnt but if we hand all of that to AI, we get this year’s deliverables out faster but lose the next decade of senior designers, the next generation of creative design thinkers, and so that still needs to be taught.
Interiors is where AI leaves the screen
The home is where AI gets out of the screen, beyond the tablet and IPhone. Woven into how light moves, how air shifts, and how sound travels. Wearables are already tracking sleep, heart rate variability and recovery. Lighting, climate and wellness systems are becoming responsive and learnable. The future is in the feedback loop between them, a lighting system that softens when a wearable flags a poor recovery score, climate control that anticipates a difficult night before the occupant has even noticed. That’s a different definition of luxury. The point is a home or building that is designed to participates in the life it houses.
Beyond the prompt
The shift I keep coming back to is the move from prompting to actual collaboration. Stop thinking of AI as something you fire questions at. Think of it as something that holds the context of a whole practice and can be asked questions across any of it. That’s a fundamentally different relationship with the technology, and a far more interesting one.
The quiet transformation here isn’t really about smart homes showing off in an entrance hall, or about photoreal renders winning attention on a feed. It’s about more prepared clients, sharper conversations, and interiors that quietly behave a little more like living systems. The aesthetic choices will still matter, they’ll just sit on top of a much more intelligent substrate underneath them. That’s the future I’m genuinely interested and excited in.
