The first working module was the engagement letter drafter, and we chose it on purpose. It is a document every fee-earner writes, it is where the client relationship formally begins, and it is exactly the kind of task that gets copied from a stale Word file with the last client's name still lurking in it. If the product could get this right, on our terms, it would prove the shape of everything that followed.

What it does

You open a new matter and drop in the essentials: the client name, the scope of work, and the fee basis. The assistant returns a full engagement letter skeleton built on the correct England and Wales template.

It is not just the opening paragraph. It is the clauses a solicitor would otherwise type from memory or hunt for in an old file — scope, fees, complaints procedure, data handling, termination. The parts that are the same every time are done. The parts that need partner judgement are laid out for you to decide.

Before this module, that document was done by hand every time: open last month's letter, search-and-replace the names, the fees and the dates, and hope nothing was missed. After it, the assistant returns the draft skeleton, and your time goes to the judgement calls instead of the boilerplate.

A draft, not a filing

We want to be precise about what comes out, because this is where a lot of legal AI oversells itself.

The output is a draft. You read it. You edit the parts that need your judgement. You sign it and you send it. The product does not send anything to your client, does not commit you to a fee arrangement, and does not treat its own output as finished. A human is on the consequential step, every time, because an engagement letter is a promise to a client and a promise is not something a machine gets to make on your behalf.

That is the pattern for the whole product: it produces a draft for review, never a filing that goes out by itself.

Why it has to know the jurisdiction

Here is the part that separates this from a general tool with a legal skin. The templates are written for England and Wales natively — drafted for this jurisdiction from the start, not machine-translated from some other market's precedent and hoped to fit.

That matters because a translated template is not a correct template. The clause structure, the regulatory expectations, the language a client and a regulator expect to see — these do not survive being run through a translation of another country's form. A draft that arrives already shaped to SRA expectations is a draft you can trust to review, rather than one you have to re-check against the wrong assumptions.

So the module is jurisdiction-aware from the first keystroke. The draft you get is built for the jurisdiction you are actually practising in, in your firm's voice, ready for the one thing only you can do — read it, judge it, and sign it.

Why it was first

The engagement letter drafter set the template, in both senses. It proved that a real, everyday document could be produced natively, kept on your machine, and handed to you as a draft for review rather than an action taken on your behalf. Every module after it is built to the same standard.

← All posts