This one is not a module. It is the rule the whole product is built around, and it is worth stating plainly because so much of the market quietly assumes the opposite.

The assistant drafts. You review, edit and sign. On anything consequential, it does not act on its own — ever.

What "consequential" means here

There is a clear line between the assistant thinking and the assistant doing. Reading a matter, summarising a document, proposing a draft — that is thinking, and it happens freely inside your browser. But the moment an action would have an effect beyond the screen, the rule changes.

Three kinds of action cross that line: writing to the matter file, sending a draft externally, and filing a court form. Each of those pauses at a pre-action gate. You see exactly what the assistant proposes to do — the specific action, on the specific matter — and you approve or reject before anything executes. Nothing crosses that line without a human hand.

Why this is a design choice, not a limitation

It would be easy to build a tool that files the form for you, sends the letter for you, saves you the click. Some tools do. The reason we do not is that a solicitor's professional judgement is not a bottleneck to be automated away — it is the thing the client is paying for and the thing the regulator holds you to. A tool that acts autonomously on a matter has quietly assumed a responsibility that is yours and cannot be delegated to software.

So the gate is not friction we regret. It is the shape of the product. The assistant does the work that is tedious and mechanical; you keep the decisions that are yours to keep.

The obligation this satisfies

This is not only good practice — it is law. The EU AI Act, at Article 14, requires that high-risk AI systems used by professionals be subject to effective human oversight: a person who can understand the output, decide whether to rely on it, and override or reject it. Our pre-action gate is that oversight, made concrete. You are never presented with a fait accompli. You are presented with a proposal.

On for everyone, from day one

Here is the part that separates us from most of the field. This oversight is not a premium feature. Many tools treat proper human-in-the-loop control — the approval cards, the audit of every decision — as something reserved for the enterprise tier, or something coming later. We think that is backwards. The smallest firm handling the most sensitive matter needs this exactly as much as the largest one, arguably more.

So the gate is on by default for every firm, from the first matter opened, at every tier. And every approve-or-reject decision is written to the audit log, per matter and per fee-earner — so the oversight is not merely claimed, it is evidenced.

That is what "built by solicitors, for solicitors" actually means in practice. Not a generic tool with a compliance page bolted on, but a product whose central design decision is to keep the solicitor in the seat — because we know the day comes when you have to account for every consequential thing that happened on a matter, and "the software did it" is not an answer any of us would want to give.

← All posts