AI is already inside most surveying workflows, whether a firm has consciously adopted it or not. Voice-to-text tools, drafting assistants, document summarisers, many are already embedded in software firms use every day. For RICS-regulated firms with one to ten staff, that creates both opportunity and professional risk, and the RICS AI Professional Standard 2026 sets out exactly what’s expected. Ignoring it isn’t a safe option: liability doesn’t shrink because an error came from a tool instead of a person.
This guide covers what the standard requires, what it means day-to-day, and how a small firm can meet it without building a compliance department to do so.
What Is the RICS AI Professional Standard?
It’s a mandatory framework, not optional guidance, it sits within the wider set of RICS professional standards and carries the same weight as any other Rules of Conduct obligation. It doesn’t ban AI. It sets the conditions under which AI can be used without compromising the client, the public, or the surveyor’s own professional standing.
Firm size doesn’t change the obligation. A sole practitioner using an AI-assisted template carries the same responsibility as a large practice running custom-built systems.
Why Should Small Firms Pay Attention?
Because small firms adopt tools fast, often without a formal review step, understandable under the pressure of practice, but it creates specific exposure:
- Liability sits with the named surveyor, not the software provider.
- AI-introduced errors, a fabricated reference, a misread defect description, aren’t always obvious on a quick read.
- Clients have no visibility into how a report was produced unless the firm tells them.
- Regulatory attention on AI use in professional services is only increasing.
None of this requires legal advice to manage. It requires knowing which tools you’re using, reviewing what they produce, and keeping a basic record of both.
What Are the Standard’s Five Principles?
Accountability. The surveyor who signs a report is accountable for everything in it, including any section AI helped draft. “The tool got it wrong” isn’t a defence.
Transparency. Clients should know when AI has contributed to their report. The standard doesn’t dictate a format, the disclosure just needs to be clear and proportionate to how much AI was actually involved.
The surveyor’s own assessment. AI can support that assessment; it can’t stand in for it. A surveyor still has to independently reach their own conclusion rather than passing along an AI-generated one as their expert opinion.
Data reliability. Anything an AI tool produces or summarises has to be checked before it reaches a client document. These systems can generate confident, plausible, and wrong output, invented regulations, misquoted standards, references that don’t exist.
Record keeping. Firms should be able to show which AI tools were used, for what, and how the output was reviewed. That’s what protects you if a regulator or a client ever asks.
Good site records aren’t a log of what was found, they’re what lets you actually write the report properly once you’re back at your desk.
That’s the record-keeping principle in practice, not a compliance exercise bolted on afterward, but part of how the report gets written correctly in the first place. This is also where a proper audit trail earns its place: not as paperwork, but as the thing that lets you reconstruct your own reasoning later.
Where Does AI Actually Show Up in Practice?
Transcription. Voice-to-text on site is low-risk, provided the output gets reviewed before it lands in a report. The risk appears when transcribed text goes straight into a client document unedited.
Drafting assistance. AI can generate draft sections fast, but every sentence still needs to be read and, where necessary, corrected. It can produce the wrong technical description or reference a standard that doesn’t apply.
Document analysis. Using AI to summarise a lease, a planning document, or maintenance records is a reasonable starting point, treat the summary as a prompt for your own review, not a verified answer. Check anything load-bearing against the source.
Defect classification from photos. Indicative only. The classification and severity in a client report have to come from inspection and experience, that’s the human in the loop requirement, and it isn’t negotiable.
What Are the Real Risks?
Four show up repeatedly for small firms:
- Fabricated information – confident-sounding content that’s factually wrong: invented clause numbers, misquoted guidance, references that don’t exist.
- Erosion of independent judgement – endorsing an AI-drafted conclusion you haven’t actually formed yourself. This is a direct breach of the standard, not a grey area.
- Liability exposure – if an AI-assisted report causes a client financial loss, responsibility sits with the surveyor. PI insurers are starting to ask about AI use, and a defensible report workflow is one of the more concrete ways to reduce that exposure.
- Misread technical context – general-purpose AI trained on general data can misinterpret surveying terminology, building pathology, or local planning nuance.
What Does a Simple Compliance Checklist Look Like?
For a firm without dedicated compliance resource, this covers the essentials:
- List every AI tool you use, including ones already embedded in your existing software.
- Check each one against the five principles above.
- Add a review step before any AI-assisted content reaches a client document.
- Add a short, plain disclosure to your terms of engagement or report template.
- Keep a basic log of which tool was used for which task, a spreadsheet is enough.
- Confirm your PI cover extends to AI-assisted work.
- Revisit your process once the standard itself is published in full.
This is also where the record-keeping principle stops being abstract. Sitarva is one example of a tool built around it directly: it timestamps every observation at capture, logs edits with the reason attached, and keeps a full record trail, it captures, organises, and preserves the evidence; the surveyor still observes, decides, and signs off. See the guardrails behind that design for the detail.
What Should You Ask Before Adopting Any AI Tool?
- Who’s accountable if this tool produces an error that reaches a client?
- What happens to the data I submit, and where does it go?
- Can I verify every output before it’s used in client-facing work?
- Does my client know this tool is in use, and would they mind if they did?
- Is this built for surveying-specific content, or is it a general-purpose model?
- Does my PI cover extend to work produced with this tool?
If you can’t answer all six with confidence, hold off on using it in client-facing work until you can.
Where Is This Heading?
AI in surveying will keep getting more capable and more embedded, that’s neither a threat worth resisting nor a shortcut worth taking uncritically. The tools that hold up are the ones handling the administrative and repetitive load, freeing surveyors to focus on inspection, judgement, and advising the client. The ones that create risk are the ones trying to substitute for that judgement outright.
The standard reflects a fairly mature position: AI has a legitimate place in practice, but only inside the same framework of accountability and competence that already governs everything else a surveyor signs their name to. Firms that build good habits now, knowing their tools, reviewing outputs, disclosing use, keeping basic records, will be in a stronger position as both the technology and the regulation move on.
Ready to see what a record-keeping system built around these principles actually looks like? Try Sitarva or take the interactive tour on the Features page.
This article is for guidance purposes only. Firms should refer to the published RICS AI Professional Standard and seek professional advice where needed.