Getting started with AI in a small law firm
By Denise Aitken, Founder of Suantraí Solutions
A useful starting point for a small law firm is to understand what people already use, agree the boundaries and choose one task that can be tested and reviewed. Assign a responsible partner or leader, involve the people doing the work and make the next decision based on what the trial reveals.
That gives the firm a practical route forward without assuming every use of AI has the same requirements.
Understand what is already happening
Ask staff which tools they use, which accounts they use and what work they use them for. Include AI features inside existing software as well as standalone assistants.
Make the first conversation about establishing the facts. Record the tool, purpose, types of information involved and who reviews the output. Do not ask people to circulate confidential examples merely to demonstrate their use.
You may discover useful practices, inconsistent checking or uses that need immediate attention. That picture should shape the firm's first actions more than a vendor's demonstration.
Agree the boundaries and ownership
Before a pilot begins, identify the approving partner or leader, the users involved, the allowed information and the reviewer. Make clear what is outside the pilot and how someone should raise a concern.
The Queensland Law Society’s Guidance Statement No. 37 addresses approval, supervision and accountability when using AI in legal practice. Consult the current guidance and court requirements applicable to your jurisdiction and intended work. This article provides an organisational starting plan, not legal advice about a particular use.
A product being described as “business-grade” or “legal AI” is a starting point for scrutiny, not a conclusion about whether the firm's proposed use is appropriate.
Choose a bounded first task
Look for a task with an identifiable source, an experienced reviewer and an output that can be checked before anyone relies on it.
| Candidate trial | Preparation to resolve first |
|---|---|
| Draft a public-facing explanation from approved firm information | Confirm source content and appoint an editor |
| Create internal training notes from a selected procedure | Confirm the procedure is current and suitable for the chosen tool |
| Produce a draft internal checklist from an approved process | Have the process owner check omissions and order |
| Explore searching a small selected knowledge collection | Review source quality, access boundaries and retrieval accuracy |
These are candidates to assess, not blanket approvals. A public-facing explanation about legal services still needs appropriate review. A knowledge-search tool needs testing to establish whether its answers are supported by the right sources.
Keep automatic communications, changes to matter records and consequential decisions outside an initial experiment unless the firm has deliberately addressed their additional requirements.
Prepare the information the pilot actually needs
Start with the selected material. Identify the current version, its owner, who can access it and how changes will be maintained.
For a precedent or knowledge collection, ask whether each item remains suitable for its intended use. A well-named outdated document is still outdated. A search tool that finds a document quickly has not established that the document is the right authority.
If important knowledge sits with one practitioner, capture and review a small useful part of it first. That work has value even before AI is involved.
Use a first-month planning sequence
The following is an example sequence, not a promise that approval or implementation can always be completed within a month.
| Period | Focus | Concrete output |
|---|---|---|
| Week 1 | Understand current use and select the task | A small usage register and pilot brief |
| Week 2 | Resolve the pilot’s conditions | Agreed tool, information boundaries, source material and reviewer |
| Week 3 | Test with controlled examples | Outputs, corrections, timings and user observations |
| Week 4 | Review the evidence | Continue, adjust or stop; record conditions for expansion |
If a prerequisite remains unresolved, change the scope or timing. The calendar should not make the risk decision.
Decide whether the work has improved
Compare the whole task, including preparation and review. Record errors and omissions, not just the time taken to generate a draft. Ask users whether the workflow is understandable and whether they can recognise when to stop and ask for help.
An encouraging result supports a decision about that task. Before expanding, check what changes when another practice area, document collection or group of users is added.
Do we need firm-wide document cleanup first?
Not for every kind of trial. Focus initially on the information the selected task requires. Broader knowledge search or connected workflows may need more extensive preparation. Our AI readiness checklist helps make that distinction.
We already use AI informally. Have we missed the starting point?
No. Your starting point is the firm as it is today. Establish what is happening, address unsuitable uses and build an agreed approach around the lessons already available.
Understand your firm before deciding what to expand
Suantraí Core helps small and mid-sized law firms examine the organisational foundations behind AI adoption. You receive a personalised AI Readiness Report with a prioritised roadmap, followed by 90 days of Decision Navigator access to explore the findings and next decisions.
AUD $650 - Founding Firm rate.
Related perspective: Should AI readiness make your firm pause?