TL;DR

Working conveyancers put AI’s usefulness almost exactly where the wider evidence keeps putting it: strong on volume and process, unreliable on anything requiring interpretation. The account comes from a roundtable convened by the legal software supplier Dye & Durham UK and written up by its marketing manager, Gayle Clarke, so read the framing accordingly — but the practitioner detail is specific and unflattering enough to be useful.

Where it earns its place

Chaired by the property law trainer Ian Quayle, the session brought together conveyancers, operations leaders and legal technologists, and the verdict divided roughly evenly between enthusiasm and caution. The enthusiasm attaches to throughput. Participants described handing over a heap of incoming client correspondence and getting something usable back inside a few minutes, where the same job had previously eaten most of a morning, and using the software to sweep files and complaints before anyone read them properly. One reported doubling the number of files she takes on.

Confidence stops where process ends and substance begins. Title analysis and points of legal interpretation still get checked by a person before anyone relies on them, because the failure mode is well known by now: output that reads fluently and confidently while being wrong, up to and including invented authorities. The liability does not move with the error — whoever relied on the tool carries it, whatever produced it.

The two second-order problems

Clients now feed the advice they have been given into ChatGPT and come back with technical queries the chatbot wrote for them, rather than ones they arrived at themselves. Separately, and more slowly, sits the training problem: entry-level work is exactly what AI absorbs first, and it is also how junior fee-earners acquire the instincts that later make them useful. Automate all of it and the next cohort arrives without them.

The practical advice was unglamorous — keep human review on substantive work, deliberately leave some routine tasks manual so juniors learn from them, and audit suppliers on how their systems actually reach an answer.

Looking forward

This lands with conveyancing under direct policy attention through the government’s AI Growth Lab, and with an AI-first firm, Keith, which raised £2m in March on a plan to strip most of the manual work out of it. It also matches what has been visible elsewhere in UK professional services this week: accountants reporting genuine hours saved, insurers warned that governance is running ahead of staff capability, and letting agents fielding AI-drafted complaints that cite law which does not exist.