7 minute read
An audit trail is the part of the RICS AI Standard most surveyors find easiest to ignore until something goes wrong. It is also the part that determines whether AI-assisted reports stand up to scrutiny from a client, an insurer, or a regulator.
The RICS Professional Standard on the Responsible Use of Artificial Intelligence in Surveying Practice took effect on 9 March 2026. It does not tell firms which AI tools to use or avoid. It does tell them what to record when AI is used, and that record is what an audit trail is for. This guide sets out what UK surveyors actually need to capture, how to capture it without adding hours of admin, and what good looks like when an audit trail is reviewed.
Why the audit trail matters more than the AI tool
The standard is technology-neutral. It does not endorse or prohibit specific AI products. What it does require is professional oversight, transparency to clients, and a documented record of how AI contributed to the work. The audit trail is the evidence that those requirements have been met.
If a complaint arises six months after a report has been signed, the audit trail is what tells the surveyor, the firm, and any third party reviewing the work, what AI did and what the surveyor decided. Without it, the surveyor is left arguing from memory. With it, the position is documented.
What a RICS AI Standard audit trail should actually contain
The standard does not prescribe a fixed format. It does set out the elements that need to be capable of being demonstrated. In practice, for a residential surveyor producing RICS Home Survey Level 2 and Level 3 reports, that means recording the following for each AI-assisted output.
“The standard does not require firms to record everything. It requires firms to record enough to demonstrate that AI was used responsibly, transparently, and under professional oversight.”
Working interpretation of the RICS AI Standard, March 2026
Five things to capture for every AI-assisted output
1. The input that produced the output
Whatever the surveyor or system passed to the AI — the prompt, the structured data, the selected defects, the photograph — should be recorded against the output it produced. The point is reconstruction. A reviewer should be able to see what the AI was asked to do, not just what it returned.
2. The output itself
The AI’s response, in its original form, before any edits the surveyor made. This is the single most important element. Without the original output, there is no way to demonstrate the difference between what the AI produced and what the surveyor signed.
3. The surveyor who reviewed it
The named individual who reviewed, edited and approved the AI-assisted content. For sole practitioners this is straightforward. For larger firms it matters more, because professional accountability under the standard sits with the surveyor who signed, not with the firm or the software.
4. The timestamp
When the AI call was made, when it was reviewed, and when the final version was signed off. This sequence is what demonstrates professional oversight in practice. A timestamp showing review happened after the AI output, not in parallel with it, is the simplest evidence that oversight was actually applied.
5. The client’s AI position
Where a client has opted out of AI-assisted writing, the audit trail should show that the opt-out was recorded and honoured. Where the client has not opted out, the standard’s transparency requirement is met by the AI-use declaration printed on the final report. Both positions should be recorded against the job.
SiteWrite
An audit trail that builds itself.
SiteWrite logs every AI call, the original output, the surveyor who reviewed it, the timestamp, and the client’s AI position automatically, against the job. The surveyor reviews, edits and signs off every paragraph, and the audit trail is built without anyone typing it.
What this means in practice
Most surveyors will not be asked to produce an audit trail at all. The work of building one is insurance, in the literal sense. When a question is raised about a report, the firms that can produce a complete record of how it was written will be in a different position to the firms that cannot.
The practical question is whether the audit trail can be built without becoming a second job in itself. Manual logging of every AI call against every job is not realistic for a surveyor producing reports under deadline. The realistic options are either a strict discipline of recording in a structured way, or software that captures the trail automatically as part of the report production workflow. For most firms, the second is the only one that survives a busy month.
For more on how the phrase library and audit trail work together, see our guide to phrase libraries for RICS surveyors.
The SiteWrite take
The RICS AI Standard audit trail is not about catching surveyors out. It is about giving them a record they can stand behind. The firms that take it seriously now will save themselves a difficult conversation later, and the surveyors who use software that builds the trail automatically will spend that saved time on the parts of the job that need their judgement.