Specification now often starts with a shortlist that an AI writes. The latest RIBA AI Report shows how quickly that has become normal. If you are weighing up what AI in architecture means for your marketing, this is the place to begin.
According to the RIBA AI Report 2026, 74% of UK practices now use AI, up from 59% in 2025. Large practices were already above 80% in 2025. This is no longer a handful of early adopters experimenting.
When an architect asks ChatGPT, Perplexity or Gemini which product suits a project, the answer usually names two to five brands. If yours is not one of them, you were not beaten. You were absent.
Architects using AI are not all generating renders. A lot of the everyday use is far more ordinary. It is research, summarising and drafting.
Common uses include:
That last group matters most to you. Product research used to mean a trawl through search results, manufacturer sites and NBS. Now it often starts with a single question and a single answer.
The shortlist that comes back shapes everything after it. Architects still check data sheets, call technical teams and consult colleagues. But they are checking the products the AI named first.
The questions are rarely "who is the best manufacturer?" They are specific, technical and tied to a project. For example:
Notice what these have in common. They name a performance requirement, a building type or a standard. They look for a product that fits, not a brand that sounds familiar.
This favours manufacturers whose content answers those questions clearly. A vague homepage will not be cited. A well-structured product page with test data, certifications and defined applications has a far better chance.
Marketing teams are used to measuring visibility in rankings, impressions and clicks. AI search for manufacturers works differently. The answer often appears in full, so the user may never visit a website at all.
That changes what you are competing for. The prize is not a click. It is a place in the answer.
There are three reasons this matters in construction.
Specification is a narrowing process. Projects start with a wide field and end with one named product. If you miss the early shortlist, you rarely get added later.
Trust is built before contact. An architect who sees your brand recommended for a specific application arrives with a degree of confidence. One who has never heard of you has to be won over from scratch.
The loss is invisible. You will not see a lost enquiry. Your analytics will show nothing, because the architect never searched for you. You simply do not appear in the reports.
You do not need a new strategy to find out where you stand. Start with a simple audit.
1. Ask the questions yourself. Put ten to fifteen realistic specifier questions into several AI tools. Use the language an architect would use, not your own product names. Record which brands appear, and whether you do.
2. Note what the AI says about you. Is the information correct? Are your certifications, product ranges and applications described accurately? Errors here point to gaps or inconsistencies in your content.
3. Look at the sources. Many AI tools show where an answer came from. Check whether they cite your site, a competitor's, a trade publication or a directory. That tells you where the authority sits.
4. Review your technical content. Can a machine read it? Data locked in image-based PDFs or behind a form is hard to find and hard to cite. Product pages should state performance, standards, applications and limitations in plain text.
5. Check consistency. Use the same product names, certification references and specifications across your website, brochures, NBS listings and press coverage. Mixed signals weaken your case.
6. Check your proof. Case studies with named project types, sectors and outcomes give AI tools and human specifiers something concrete to work with.
Often the fixes are less about technology and more about clarity. Good website design puts technical information where people and AI tools can find it. Well-structured technical brochures keep that information consistent across every touchpoint.
The RIBA AI Report shows where the profession is heading. Adoption is already high, and it is rising fast. The manufacturers who treat AI search as a specification channel now will be easier to find as it grows.
Start by looking. Ask the questions an architect would ask and see who comes back.
Want to know if AI recommends your products? Talk to us.
Trusted by 23+ clients
