RICS' New AI Guidance: Who Checks What Your Construction Firm Sends?

A handover record to use with the workflow

Field	What to record
Owner	Person responsible for the final statement
Source	Approved facts, document and version
Check	Technical or commercial claims requiring confirmation
Approval	Who approved the final wording and imagery

This is Market Maestro's content handover suggestion, separate from RICS's own guidance.

Someone in your business puts a few project notes into AI. A minute later, there is a polished case study ready for the website, a tender response or LinkedIn.

The wording is better. But has the story changed?

Perhaps "we installed the flooring" has become "we delivered the refurbishment". A target completion date has become a confirmed result. A project value has quietly become the value of your package.

Those are hypothetical examples, but they show the check that matters: can someone in your business stand behind every claim before it leaves the business?

What has RICS announced?

On 19 October 2026, RICS published an insight paper called How to build a judgement layer for AI use in surveying. It explores how surveyors can distinguish AI-generated output from professional advice and decide how far that output can be relied upon.

The paper builds on RICS' existing Responsible use of AI in surveying practice professional standard, which came into effect for members and regulated firms on 9 March 2026.

This is a new insight paper, not a new rule applying to every construction company. For contractors outside RICS regulation, it still raises a useful operational question: where does a named person check the AI-assisted work?

Read the RICS announcement and insight paper and the professional standard.

A practical version for a busy construction firm

The following is our suggested workflow for project evidence and marketing. It is not RICS' framework or a substitute for the professional standard.

You do not need another meeting for every LinkedIn post. You need a clear source, a sensible review and someone who owns the final version.

1. Give AI a source pack you are allowed to use

Start with approved project information: your scope, relevant dates, photographs cleared for publication and any verified results.

Keep the distinction between the main contract and your package explicit. Write down whether the job is complete, ongoing or at a particular installation stage. Identify which client names and figures can be published.

Use your firm's approved AI environment and data policy. Permission to work on a project is not automatically permission to upload its tender documents or publish its details.

If a fact is missing, the draft should say it needs checking. It should not fill the gap with something plausible.

2. Check the claims against the evidence

A fluent paragraph is harder to challenge than a short list of facts. Pull the important claims out of the draft and compare them with the source.

Check	Question to ask
Scope	Does this describe our actual contribution?
Figures	Is this our package value, the whole project value or an estimate?
Results	Is the outcome measured, reported by someone else or still a target?
Dates	Does the wording match the current construction stage?
Attribution	Are the client, contractor and other specialists credited accurately?

The person who knows the job should review those facts. Asking AI whether its own answer is accurate does not replace opening the source.

3. Separate a capability claim from a promise

"We have delivered this type of installation" is different from "we can complete your installation in six weeks".

Check for statements about availability, programme, pricing, warranties, compliance or performance. Those need approval from whoever has authority to make that commitment.

A marketing draft should not accidentally become a commercial offer. A tender response needs the bid owner's review against the actual question and submission requirements, with the relevant technical and commercial checks.

4. Approve what will actually be sent

Check the final document or scheduled post, including the photograph and caption. Reviewing an earlier draft is little help if someone later adds an unsupported headline.

Record who approved it and retain the source material. For a simple post, that might be a dated approval in your existing project folder. More consequential work needs the review process appropriate to that work.

Follow any client, procurement or professional requirements about declaring AI use. Do not assume the same rule applies to every document.

Why this matters for winning work

Your project content should make it easier for a prospective buyer to understand what you do, where you work and why your experience is relevant.

An exaggerated case study can create the wrong conversation. Accurate evidence gives your team something useful to send when a main contractor asks whether you have delivered a similar package.

That is also how it fits into LinkedIn business development. The evidence supports your profile, your posts and your conversations with relevant buyers. Publishing more frequently is only one part of becoming familiar to the people you want to work with.

For help choosing a tool, use our ChatGPT versus Claude comparison for construction companies. For getting started with Claude, use the first-hour guide. This review process comes after either tool has produced its draft.

One useful change to make this week

Take one recent case study or tender answer and ask the project lead to check its scope, figures, results, dates and attribution against the source documents.

Then make that check part of the approval process for the next one.

If your project evidence is strong but the right contractors still do not know you, see how our LinkedIn service works. We help construction businesses identify relevant buyers, strengthen director profiles, publish credible evidence and build familiarity through targeted outreach. You can also read the Bayliss Decorators case study or send Jude an enquiry.