Last week’s data settled whether agentic AI can transform legal work. The harder question it exposed is why some teams reorganise around that capability in months while others leave it on the shelf. Ed Molyneux on the capability-adoption gap, and why conveyancing’s capacity crunch makes it the sector with the most to gain and the least room to wait.
For months the honest doubt about agentic AI was whether the capability was real. The OpenAI Codex data reviewed in this column last week put that doubt to bed: inside a low-friction organisation, agentic tools went from novelty to near-total share of output in under a year, and the functions that started late moved fastest.
That retires one question and raises a sharper one. If the capability is genuinely available, why do some teams absorb it in a quarter and others not at all? The distance between what the tools can do and what a firm actually does with them is now the whole contest, and it has surprisingly little to do with the technology.
Behind the headlines
The most instructive figure in that paper is not the headline about engineers. It is that inside OpenAI, where every function had identical tools, access and permissions, legal usage sat near zero in January and reached roughly three quarters of output within months, while other functions stayed low for far longer. Same technology, same building; the only variable was how quickly each team rebuilt its work around delegation and review.
The paper reaches for an old analogy to explain it. Factories bought electric motors decades before they saw the productivity gains, because the gains only arrived once they stopped bolting the new motor onto the old central drive shaft and re-laid the whole floor around it. The constraint was never the motor. It was the willingness to reorganise.
That should concentrate minds in conveyancing more than in most sectors, because ours carries a structural capacity problem a software team does not. Transaction volumes swing violently with the market, and firms cannot flex headcount to match; an experienced conveyancer takes years to train and cannot be hired into a spike. Liability and judgement concentrate on a shrinking number of senior people, and the profession has spent a decade being asked to handle more matters with fewer hands.
In that setting the capability-adoption gap is not an abstraction on a chart. It is the instructions a firm had to turn away, the search read too quickly under load, the enquiry a junior answered because the person who should have was not free. Conveyancing has the most to gain from closing the gap and the least room to sit on it.
The choice is yours
Closing it, though, is not a purchasing decision, and this is where firms go wrong. Buying the cleverest model changes very little if the work around it stays the same.
Three things actually move the needle, and none of them is a product.
The first is to find where senior judgement is being spent on production that could be delegated, and to redesign those tasks around a person who directs and checks rather than types.
The second is to make sure the facts an agent relies on are reachable and trustworthy, because an agent reasoning over a mis-keyed lease term or an unverified search result will be confidently, plausibly wrong.
The third is to insist that its reasoning can be tested against rules a qualified lawyer has actually signed off, rather than taken on trust. Trusted inputs and verified judgement are two different layers, and a firm needs both; the model sits on top and is the least of it.
The reassurance in the same evidence is that late starters move faster than the pioneers, because the playbooks already exist by the time they begin. No conveyancing firm is behind on capability. What remains is a choice about how quickly to reorganise, and the firms that treat it as a choice will pull away from those waiting for permission.
The infrastructure that makes that possible – trusted and provenanced property data agents can reach and reasoning that can be verified against signed-off rules—is the work we are engaged in at Moverly, on open foundations others can inspect. The tools are no longer the hard part. Deciding to rebuild the work around them is.
About the author
Ed Molyneux is co-founder and CTO of Moverly, the property intelligence platform working with LMS and Connells Group to bring structured, verified data to property transactions. He is the architect of the Property Data Trust Framework (PDTF), the open standard for machine-readable property data now being adopted across the industry. Ed writes about AI, property data infrastructure, and the future of conveyancing.
