From 1 November 2026, a Scottish practice unit needs one central record of every trust where the firm or its people act as trustee — all of them, not just the new ones, and produceable within 21 days of being asked. Here is what it has to contain, and how to build it without a scramble.
Most of the conversation about this rule has been about the deadline. The more useful thing to understand is the shape of the work, because that is what decides whether this is a fortnight of effort or a very long autumn.
The short version: the information is almost certainly all there already. It is sitting in files, in styles, in old correspondence and in the heads of people who have been at the firm a while. What does not exist yet is one place where it can be read from, and a way of keeping it right afterwards.
Rule B6.24 requires a practice unit to keep a central record of trusts. It bites where a regulated person who is a manager or an employee of the practice unit acts as a trustee, and where a body owned or controlled by the practice unit acts as trustee — the trustee company arrangement that a great many firms use.
Two details matter more than the rest, and they are the ones that catch people out.
It applies whether or not the trust needs to be registered with the Trust Registration Service, so TRS registration is not a proxy for scope. Client account money held in trust is outside it, and so are executries during administration — though if an executry continues beyond that point as a trust in its own right, it comes back into scope.
The list is longer than a name and a date, which is the main reason this takes real work rather than an afternoon.
The information exists. What does not exist yet is one place to read it from, and a habit that keeps it right afterwards.
Three passes, in this order. The order matters, because doing it the other way round is how firms end up doing the hard part twice.
First, find the population. Before anybody opens a file, work out how many trusts you are actually talking about and where they live. Trustee company appointments, partner appointments, dormant matters, closed files that were never quite closed. Most firms find the number is either much smaller or much larger than the guess, and either answer changes the plan.
Second, take what you already hold. A good deal of the record can be assembled from what is already in your system without anybody reading a file — names, dates, references, and often the settlor. Do that sweep first so the manual pass is only filling gaps rather than starting from nothing.
Third, work the gaps. What is left is mostly historical trustees, asset detail and the location of deeds — the genuinely manual part, where somebody has to look. By this stage it is a short list of known questions against known files, which is a very different job from “go through the trusts”.
This is a good example of a job with a clean line through the middle of it.
The reading is the part worth handing over. Pulling names, dates, settlors, appointments and asset descriptions out of years of scanned deeds, minutes and correspondence is exactly the sort of fetching that current AI does well and quickly, and it does not get bored on file two hundred. It can also flag the awkward cases — a trust with no obvious termination date, an appointment that never seems to have been minuted — and put them in a pile for a person.
The judgement is not. Whether a matter is in scope, how a trust is classified, and whether what has been found is actually right are all questions for a solicitor, and the record is signed off by a person either way. Built properly, every entry shows which document it came from, so checking it is quick rather than a second full pass.
The record is not a project with an end. The part firms find hardest a year later is that new trusts, new appointments and changes of trustee all have to reach it, which means it has to sit inside how the work already happens rather than beside it in a spreadsheet somebody remembers on a Friday.
That is worth designing on day one, when it is a small decision, rather than after the first time somebody notices the record has drifted.
The same reading-and-drafting split, done properly, is set out under doing it safely. Two related pieces: residual client balances, worked through the same way, and three things AI should never do in a cashroom. If you want a hand sizing this properly against your own trust population, that is what the audit is for.
This is a plain-English summary written to be useful, not legal advice, and it is not a substitute for the rule itself. Read Rule B6.24 as published by the Law Society of Scotland, and take your own view on scope. If anything here does not match the rule as published, the rule is right and we are not.
Twenty minutes on the phone, free, no obligation. Tell us roughly how your trusts are held and we will tell you honestly what the work looks like — including if the answer is that you can do it yourselves.