
By Denise Aitken, Founder of Suantraí Solutions
The most useful place to start with AI is a task you understand well enough to judge. Choose something that takes time, happens regularly and produces a result someone can check. Then establish what information the tool may use and test whether it actually improves the work.
You do not need every process documented or every folder reorganised before you begin. You do need to understand the part of the business your first experiment will touch.
Look at the work your team repeats during an ordinary week. Where do people rewrite similar text, search for the same information or turn rough notes into something usable?
Write down three candidates. For each, ask who does it, how often it happens, what makes it difficult and what a good result looks like. “Use AI in marketing” is too broad to test. “Turn an approved service description into a draft email for an existing customer group” is much easier to assess.
A useful first task usually has a clear beginning and end, an available reviewer and limited consequences if the draft is wrong. Work involving consequential decisions, confidential records or automatic external actions needs more preparation.
Ask the person doing the task to walk you through one example. Note where the information comes from, which version they trust, what they change and who checks it.
This can reveal a simpler answer than buying AI. If the real problem is that nobody can find the current template, fixing the template location may help immediately. If the job follows an exact rule every time, conventional automation may be a better fit. AI becomes more useful when you can explain what part of the task needs interpretation, drafting or summarisation.
For an example of that preparation, read The AI Was the Easy Part.
Give the person running the experiment a short written brief:
Use fictional or public material for the first test where possible. Do not assume that a paid subscription automatically makes a tool suitable for personal, confidential or commercially sensitive information. Check the specific plan, terms and settings for your intended use.
Our AI readiness checklist helps you identify what needs attention for the task you have chosen.
Keep a small record of the current process and the AI-assisted version. Measure preparation time, time spent using the tool, checking time and corrections. Also record whether the finished work meets the same quality standard.
For example, imagine a task normally takes 30 minutes. AI produces a draft in three minutes, but preparing the input takes five minutes and correcting the draft takes 15. The saving is seven minutes, not 27. That may still be worthwhile, but it is a different buying decision.
This is an illustrative calculation, not a promised result.
At the end of the trial, choose one of three actions:
Expand only after you understand why the first trial worked. A good drafting experiment does not establish that the same tool should search every business document or send messages without review.
A limited experiment needs a clear purpose and boundaries. Wider adoption needs decisions about ownership, information, training, spending and how tools fit together. You can develop those decisions while learning from a carefully scoped trial.
First check whether your existing software can support the chosen task. Then compare suitable options using the same test material. Our guide to choosing AI tools explains the questions to ask before buying.
If you run a small or mid-sized law firm and want to understand what can start now and which foundations need attention, Suantraí Core provides a structured next step. It combines an organisational assessment, a personalised AI Readiness Report and 90 days of Decision Navigator access.
AUD $650 - Founding Firm rate.
See what the assessment includes · View sample report findings
Further reading: Australian Government guidance on AI in business.