Five questions to ask AI vendors about the new RICS AI Standard
Co-founder, Tendermark
If you are a building surveyor looking at AI tools for tender comparison, defects analysis, schedule extraction, or anything else where an AI output ends up shaping your professional advice then the new RICS AI Standard changes how you should evaluate your tooling vendors.
The Responsible Use of Artificial Intelligence in Surveying Practice standard came into effect on the 9th of March 2026 and is mandatory for every RICS member and every RICS-regulated firm. It applies whenever an AI output has a "material impact" on the service the client receives (which probably covers most of the AI tools currently being sold to surveyors).
We are building Tendermark (an AI-assisted tender comparison tool) to fit inside this standard from day one. And whilst I'm not a surveyor, and nothing in this post is professional advice on how to comply, I have spent enough time with practitioners and the RICS AI Standard to spot the key questions a surveyor should be asking the people selling them software — and I've also noticed that very few vendors are providing good answers.
Here are the five questions I think you should be asking:
1. Do you claim your tool is "RICS-compliant"?
If they say yes, that is your first warning.
Compliance attaches to the surveyor and the firm. It does not attach to the software.
The standard is not a certification regime. RICS does not approve or certify tools. It sets professional obligations on you — to do due diligence, to document risks, to disclose to the client, to make professional judgement yourself rather than outsource it to software. Any vendor claiming their tool is "RICS-compliant" is overclaiming, and that overclaim should make you wonder what else they have not read carefully.
The honest framing — and the one to look for — is that a tool is built to support the surveyor in meeting their obligations, not just speed up their workflows. That is a meaningful distinction. Vendors who get it right have read the standard. Vendors who get it wrong are guessing.
2. Do you supply a template for the clause 1.2 determination?
Clause 1.2 of the standard requires the surveyor or firm to record, in writing, that they have determined the AI use to be material — and to capture the reasoning. It is a short note: what the tool does, why its output is material, the reasoning behind the call.
This sits before every other obligation in the standard. Without it, you cannot demonstrate to a regulator that you have engaged with the standard at all. And yet most AI vendors I have looked at do not mention it exists. They will sell you a tool, take the money, and leave you to discover months later that there is foundational paperwork you were supposed to have on file from day one.
A vendor who has thought about this provides a template. You read it, edit it, and own it (but the work of structuring it has been done for you).
3. What does the interface actually say at the moment of decision?
The standard is explicit that AI must support professional judgement, not replace it. So when you are demoing a tool, keep an eye on the language being used.
A tool that says "we recommend Contractor B" has done a piece of professional judgement on your behalf. That is the wrong direction.
A tool aligned with the standard surfaces things — flags an unusual price, notes a missing return, highlights a discrepancy — and leaves the judgement with the professional (where it belongs). The verb the tool uses is "flag." The verb you use is "decide." If a vendor's interface uses the language of recommendation, they have either not read the standard or have decided to ignore the part that matters most.
4. Can you show me, on demand, what the AI extracted versus what I changed?
Clauses 4.1 and 4.4 require you to be able to explain the tool, in plain English, to a regulator or a client who asks. That includes the due diligence you did before adopting it — and, by implication, what the tool actually did during a piece of work. The standard does not prescribe a specific format; it requires that you can demonstrate and explain, on request.
When the work happens in spreadsheets, the audit trail is the spreadsheet — you can see every cell, every formula, every edit. When it happens in an AI tool, the question is whether the tool preserves the same transparency. If the AI extracted a price and you overrode it, is the original extraction still visible? If you corrected a misread, is there a record of what was changed and why?
A vendor who has thought about this preserves the original AI extraction alongside every correction, and surfaces both in the interface — so that at any point during or after the work, you can see exactly what the AI produced and what professional judgement was applied on top of it. A vendor who hasn't will tell you "the system keeps track of everything" and not be able to show you what that means in practice.
5. Do you supply suggested terms-of-engagement language?
Clause 4.3 requires you to tell the client, in writing, that AI is being used in the delivery of their service. It is disclosure, not consent — the client does not have to sign anything; they have to be told.
In practice this can sit at the firm level (a single line in your standard terms of engagement that covers every client at once) or at the project level (an email or letter for a one-off engagement). Either is fine. Neither is hard to do.
But it is paperwork your firm has to produce. A vendor who has thought about the standard supplies suggested language you can adapt. A vendor who hasn't will leave you to draft it yourself, which is a small thing — until you multiply it by every AI tool your firm adopts.
What we built
Tendermark is built to give honest answers to all five of these questions. We never claim the tool is "RICS-compliant" — we describe what it does to support a surveyor in meeting their obligations. We supply a clause 1.2 determination template. The interface uses "flagged for your review" everywhere, never "we recommend." The original AI extraction is preserved immutably alongside every correction — so at any point you can see exactly what the AI produced, what you changed, and why. And we ship suggested terms-of-engagement language alongside the product.
All of this lives in a free compliance pack you can download here — no email required. Whether or not Tendermark is the right tool for your firm, the pack is genuinely useful as a reference. Use it to evaluate any AI vendor you are looking at, including us.
I want to be honest about the limit of my view here. I am not a surveyor. The questions above are the ones that struck me as a careful outsider reading the standard, but surveyors will spot questions I have missed. If you are evaluating AI tools and have thoughts on what a vendor should be asked beyond these five, I would genuinely like to hear from you.
The standard itself is available from RICS. It is twenty pages and well written. Worth an hour of anyone's time.
Alexander Wingfield is co-founder and CEO of Tendermark. Tendermark is an AI tender comparison tool for building surveyors, built to the RICS AI Standard from day one.