There is a lot of noise around AI in legal practice.

Some of it is genuine excitement. Some of it is fear. A fair amount of it is people who have never run a law firm explaining what law firms should do.

Here is a more useful frame. After working with professional services firms on AI implementation, the pattern is clear. The firms getting genuine value from AI are doing very specific, boring things. And they are deliberately not doing the exciting things that most of the commentary focuses on.

What the NSW courts actually say

Before any discussion of AI in law, it is worth understanding the regulatory context.

The NSW Supreme Court's Practice Note SC Gen 23, which governs the use of generative AI in proceedings, commenced on 3 February 2025. Similar notes followed in the District and Land and Environment Courts.

The position is not anti-AI. It is anti-unchecked AI. Practitioners who use AI to assist with court documents must disclose that use. They remain fully responsible for the accuracy of everything they file.

The Law Society of NSW has published guidance on responsible AI use in legal practice. The message is consistent: AI as a tool, practitioner as the accountable professional.

This is the right frame. Not "can we use AI" but "where does AI help without creating risk."

The high-risk uses (and why smart firms avoid them)

Legal research is the area that attracts the most AI interest. And it is also the area with the most cautionary examples.

In 2024, a lawyer was referred to the NSW Legal Services Commissioner after filing AI-generated citations that did not exist. The case became a prominent warning about the limits of AI in legal research contexts.

The lesson is not that AI cannot assist with research. It is that any AI output that gets filed, cited, or relied on in a matter must be reviewed by someone qualified to catch what the AI got wrong.

Firms that use AI for legal research and treat the output as finished work are taking a risk that the regulatory framework makes very clear.

Where the value actually is

The genuine value in AI for law firms right now is in the parts of the week that have nothing to do with legal work.

Client intake. The process of collecting identity documents, matter details, and source of funds information from a new client is time-consuming, repetitive, and identical every time. A well-built intake workflow sends a single link, collects everything, and presents the partner with a complete file before the first meeting.

Time capture. Partners lose billable time constantly to the gap between doing the work and recording it. A voice-based time capture tool lets a partner note a time entry in 10 seconds while still in the moment. The AI formats it, assigns it to the right matter, and logs it.

Scheduling. Any matter involving multiple parties, expert witnesses, or mediators involves a scheduling problem. Coordinating five calendars by email is a multi-day exercise. A shared availability link resolves it in minutes.

Document summarisation. Partners routinely need to work through large volumes of financial records, correspondence, or file material. AI summarisation of these documents is fast, accurate for the purpose of orientation, and frees the partner to focus on the legal analysis rather than the document triage.

These are not exciting use cases. They are also not risky ones. And collectively, they return a significant amount of time to the people in a firm who cost the most per hour.

The confidentiality question

Every law firm asks the same question when AI comes up: what happens to client data?

It is the right question.

Cloud-based AI tools process data on external servers. For most client information in a law firm context, that is a problem. Client privilege, confidentiality obligations, and the sensitivity of the matters that come through a law firm's door all point to one answer: client data should not leave the firm's systems.

The tools that solve the admin problems above can be implemented with client data remaining local. Meeting summaries processed on a local device. Intake information stored on the firm's own infrastructure. Document summarisation running on a private instance.

This is not a barrier to using AI. It is a specification for how to use it correctly.

The practical starting point

For a law firm looking at AI for the first time, the most useful first move is not to pick a tool. It is to map the week.

Where is the non-billable time going? Who is spending it? How often does the same task repeat? What is the cost of each hour spent there?

When you have that picture, the right tools become obvious. And the cost of doing nothing becomes a number you can actually see.

VerusLink helps professional services firms audit, optimise, and activate the admin that's draining their highest-value people. Free AI audit at veruslink.au/audit.