AI in practice

The solo lawyer and AI

Where it really saves time, and where it does not.

[date] · [author] · 6 min read

A lawyer who practises alone is researcher, drafter, secretary and bookkeeper at once. So AI looks like a tempting promise: hours back every week. The promise is true in places and false in others. The difference is not in the tool but in one question: how long does it take to check its work?

One rule

AI saves you time when checking its work is faster than doing the work yourself. When checking takes longer, or cannot be done, it saves nothing. It moves time from writing to reviewing, and adds risk.

Where it really saves time

  • Reaching the governing text: from a question in your own words to the article that governs it, as long as the article itself appears beside the answer.
  • First drafts of routine papers: a formal notice, a demand letter, a reply to a notice. You know what you want to say; the tool writes the structure; you review.
  • Reading long documents: summarising a sixty page contract, or pulling the dates, amounts and obligations out of a case file.
  • Ordering the facts: turning scattered notes into a timeline you can check in minutes.
  • Translation drafts: a starting point for a contract in English, as long as you check the terms yourself.

Where it does not

  • When you cannot see the source. An answer without an article you can open means you will look for the article from scratch.
  • Anything that changed recently. New amendments are the weakest point of any general model.
  • Numbers: procedural deadlines, limitation periods, court fees. One wrong number can cost a right.
  • Judgement: what to argue, what to leave out, how to talk to your client. That is your work, and it cannot be handed off.
  • Client secrets in general tools. Pasting a client's file into a tool when you do not know where it keeps the file, or whether it trains on it, puts you up against the data protection law and your duty of professional secrecy.

Why a general model gets this wrong

A general model is good at form, because form repeats across everything it has read: the structure of a notice, the order of a memo, the language of correspondence. It is weak on Egyptian detail, because that detail is rare in its training text: an article number, a limitation period, a particular ruling. The model learns from repetition. What it has seen thousands of times, it masters; what it has seen a few times, it guesses. That is the trap: polished form makes a guess look like knowledge. Use it for form, and take facts and texts from the source.

In Bayan

Bayan is built to make checking fast: the article opens beside the answer, and a ruling opens with its full citation. You get AI's speed at writing without losing time hunting for a source you never saw.

Book a team demo
Share

The month's key legal updates and what's new in Bayan, once a month.

Ask Bayan your first question.

Or book a team demo and bring a real contract: thirty minutes is enough to know whether Bayan fits your practice.

Book a team demo