Architects
Your last project photographed beautifully. The question is whetheran AI assistant can actually read itwhen a client asks who to hire.
A prospective client rarely starts with your website anymore. They start by asking ChatGPT or Gemini to shortlist firms for a specific brief — a barn conversion, a Passivhaus retrofit, a heritage-listed extension — and the assistant answers from whatever text it can actually parse about you, not from the portfolio images you spent the most time laying out.
The client comparing firms for their project has already asked an AI which ones fit before they ever open your site — if it can't tell what you've built or what you're licensed to build, you don't make the list.
Architecture firms sell trust and specialization more than any single service line, and both currently live inside PDFs, image carousels, and awards pages that AI systems struggle to parse. Your RIBA stage experience, your planning track record in a specific borough, the fact that you've delivered three actual Passivhaus certifications rather than just expressing interest in sustainability — none of it reaches an AI assistant unless it exists as retrievable, quotable text. Clients researching an architect are unusually risk-averse: they're committing months of fees and a project they'll live in for decades, so they lean on AI to pre-filter by specialization, project type, and planning-approval success before they ever pick up the phone. A firm that only exists online as a gallery of finished buildings is invisible at exactly the research stage that decides who gets the enquiry call.
How your customers ask
How clients actually ask
Why it matters
Where firms lose the answer
The portfolio is a gallery, not a record
Your site shows a finished barn conversion as ten beautiful photos and a project name. It doesn't say what the planning constraint was, which local authority approved it, or what stage of RIBA work you delivered. An AI assistant can't extract a specialization from an image, so it defaults to whatever competitor wrote the same project up in words.
Specialization gets flattened to "residential"
Most firms describe themselves broadly to avoid turning away work, but a client asking about a specific building type or planning situation needs the opposite: a narrow, provable match. If your site never states "heritage extensions in conservation areas" in plain text, an AI has no reason to surface you for that exact brief over a generalist competitor.
Licensing and registration live off-site
Chartered status with the ARB or RIBA membership is often mentioned once, in small print, or not at all on the working pages of the site — assumed as background credibility rather than stated as a fact. Clients (and the AI systems answering them) increasingly treat licensing status as a trust filter before price ever comes up, especially for anything involving structural work or planning risk.
Go deeper
Articles for Architects
Fundamentals
AI Visibility for Architects: Why Clients Have to Find You in ChatGPT
Practice
Image-Heavy Portfolio, Invisible to AI: How to Make Your Projects Readable
Strategy
Architect Near Me: How to Show Up in Regional AI Answers
Strategy
Timber Construction, Heritage Protection, Passive Houses: Why Specialization Is Your AI Signal
Data & studies
What Clients Really Ask AI Systems About Architects: A Data Analysis
Questions firm owners actually have
My portfolio is almost entirely images. Do I need to redo it?
No, but you need to add what the images can't say. Keep the photography and add a written paragraph per project covering building type, site constraint, planning authority, and what stage of work you delivered — that text is what an AI assistant can actually retrieve and quote back to a client.
Is it risky to claim a specialization if we also take general residential work?
State the specialization as what you're known for, not as the only thing you do. Clients searching for a specific project type respond to specificity; general residential work still comes through referral and reputation the way it always has, so naming your niche doesn't cost you the broader pipeline.
Does chartered status with ARB or RIBA actually affect how AI systems describe us?
It affects whether a cautious client trusts the answer enough to act on it. Registration status is exactly the kind of fact-checkable, low-ambiguity detail that AI systems can state with confidence, so making it easy to find and unambiguous in your own text removes one more reason for a client to hesitate or ask a follow-up question elsewhere.