Nearly everyone has used an AI chatbot, you ask a question and almost instantly you get an answer. Some of you may have taken it a step further and asked the chatbot to summarise a document or email. But fewer people have ever used an agent, well, at least consciously, and I’ll explain that part in a bit.
What an AI agent actually does
Firstly, what does an AI agent do? Put simply it takes continuous, independent actions, uses software tools and adapts based on results. An agent is given a goal and works towards it.
Instead of one prompt and one reply, an agent takes multiple steps on its own. It can open documents, search a folder, pull figures into a spreadsheet, cross-check them against a contract and draft the output, repeating until the task is done.
Metroun LearningShort courses with structured CPD, 20 minutes eachStart ›You are probably already using one
You might already be using agents. If you’ve ever asked an AI chatbot to search the web and find you specific information, or recommend a product, that is an agent at work. If you’re a beginner, setting one up is actually quite simple. Whether you use ChatGPT or Claude, you’ll notice something called connectors. These connectors allow the AI to talk to different systems you have access to. For example, Microsoft 365. As soon as it’s connected you can ask your AI to summarise emails, or draft replies. It can do this while also taking into consideration documents you have saved on SharePoint or OneDrive, like a programme, contract or design documents. Suddenly your chatbot has become an agent. It doesn’t just stop there though, connectors are now becoming available for financial systems, site record systems and many more. If you would like us to put together a tutorial on how to set this up, let us know. In the meantime, just ask AI.
Remember this. A chatbot talks. An agent does.
Where a QS would genuinely use one
So, where would a QS genuinely use one? Here are some examples:
- Application reviews, checking a subcontractor’s application against the measured quantities and flagging the differences before you assess it.
- Contract interrogation, pulling every clause relating to payment dates, notices or compensation events out of a 300-page document in seconds.
- Negotiating claims with subcontractors and clients.
Where it can catch you out
Although, it’s worth making sure you don’t get caught out. Look at what has happened in the legal profession. UK courts and tribunals have now dealt with dozens of cases involving fabricated or miscited authorities produced by AI, and in April 2026 a judge referred two solicitors to the SRA over it.
The lesson transfers directly. An agent will state a wrong payment date in exactly the same confident tone as a right one, and if that date ends up in a payment notice, the consequences are yours to own. Use AI for first drafts, not final ones.
What the RICS standard requires
The RICS professional standard on the responsible use of AI in surveying practice came into force in March 2026. It is mandatory for RICS members and regulated firms wherever AI output has a material impact on the service.
An AI agent is an assistant, not a surveyor. RICS has put your professional judgement, not the AI’s output, at the centre of this standard. The QSs who do well will not be the ones who avoid the technology, or the ones who trust it blindly, they will be the ones who understand how it works, and manage the risks.
Find out how we can help you at metroun.co.uk
