The right template, the wrong problem: designing professional judgement into your systems and AI
A client recently complained that their lawyer never read the brief, and then implied they had. It is a familiar story, and AI could make it more common, not less. The answer is not to avoid technology, but to design judgement into it.
Output is cheap. Judgement is not.
Researcher and former lawyer Chantal McNaught summed up the risk in one line: lawyers can deliver the right template, the right citation, and the wrong problem. Technically correct work that misses what the client actually needed.
It is not a new problem, but technology can make it worse. Practice management systems reward what they can measure: time recorded, documents produced, tasks closed. Workflows are often built around steps rather than outcomes. And AI now makes a first draft, a summary or a research note almost free. When output costs nothing, it is easy to produce more of it and think less about whether it answers the question.
Clients notice. Firms are already receiving what one commentator called “AI slop grenades”: long, unreviewed AI-generated material that lands on someone else’s desk. Nobody wants their own firm to be the one throwing them.
What professional judgement involves
Drawing on a definition by Justice Lee, Chantal McNaught describes professional judgement in four parts. Each one says something about how systems and AI should be designed.
| Professional judgement means… | What your systems and AI should do |
|---|---|
| Understanding the facts that matter, and communicating them | Capture the client’s objective at the start, in their words, and keep it on the matter where everyone can see it. Use AI to summarise where a matter stands, with sources, so the lawyer starts from the facts rather than a blank page. |
| Separating what is important from what is merely available | AI can surface everything. The lawyer decides what matters. Design summaries around “what has changed and what matters”, not “everything we found”, and never send a client the raw output. |
| Appreciating the consequences, by asking and researching | Key dates, risk flags, precedent notes and checklists bring the right questions forward at the right moment. AI can do the research legwork; weighing the consequences stays with the lawyer. |
| Helping the client understand the decision | Reporting that explains the options and the recommendation in plain language, and time protected for the conversation that goes with it. |
In every row the pattern is the same: technology does the finding, sorting, chasing and first drafting, so the lawyer has more time for understanding and explaining. Done badly, it produces more output, faster, with less thinking behind it.
Five ways to design judgement in
- Record the client’s objective at matter opening. Make it a required field, in plain words: what does the client need to happen? Show it at the top of the matter, and in every AI summary.
- Build workflows around outcomes, not activity. Start from the result each type of matter must achieve, then design the steps. Our workflows insight and free workflow capture tool show how.
- Put review where the judgement happens. Supervision and file reviews often check compliance. Add one question that checks judgement: does the advice answer the client’s actual objective?
- Use AI for volume, with checking built in. AI reads, sorts and drafts; a named person checks before anything reaches the client. Let AI read before it is allowed to act, and keep a record of what it did (see Choose your AI gatekeeper).
- Measure more than volume. Alongside time and fees, track what clients value: time to a clear answer, how often advice has to be revisited, and what clients say at the end of the matter.
Technology should buy time for thinking
The best reason to invest in systems and AI is not to produce more. It is to give lawyers back the time that professional judgement needs, and to make sure that time is spent on the client’s real problem. Firms that design for that will deliver better advice with less effort. Firms that do not will deliver more templates, faster, to the wrong problem.
We help firms design workflows, systems and AI around the outcomes their clients need, independently of any vendor.
Prompted by a LinkedIn post by Chantal McNaught on professional judgement (October 2026), including her adaptation of Justice Lee’s definition, and the discussion that followed. The design recommendations are our own. If you think anything here is out of date, please tell us and we will correct it.
Where does your firm stand?
Our free practice management review shows how much of your current system you really use and, if you are thinking of moving, builds an independent requirements specification that takes your reasons into account. Nothing is sent to us.
Start the free review Talk to us