Recent advice from Treasury has revealed Australia’s AI challenge - around two thirds of businesses are using the technology in some form, but fewer than one in 10 say they are using it significantly.
Treasury describes this as “widespread but shallow” adoption, and warns that giving businesses access to AI will not be enough to lift productivity. The real gains will come when they use it to change their processes and the way work gets done.
That can sound daunting, particularly for a small business without a large technology team or the time to test every new AI tool or LLM (large language model). But one of the clearest places to start is in the conversations every business is already having.
Think about what happens after a meeting. A decision has been made, responsibilities have been agreed and everyone leaves knowing what needs to happen next. Then someone has to write the recap, assign the actions, update the relevant systems, brief the people who were not there and make sure everything is completed before the deadline.
The meeting moved the business forward, but it also created another layer of work.
Customer conversations follow the same pattern. A customer explains a problem or tells a business what they need, but once the call ends, an employee still has to record the details, update the customer’s file, contact another team and arrange the next step. The conversation contains the information needed to move forward, yet people still spend hours manually moving that information from one place to another.
This is a practical example of what shallow AI adoption can look like. A business may use AI to produce a transcript or summary, but if an employee still has to complete every step that follows, the underlying process has barely changed.
What happens after the summary?
There is no doubt that capturing conversations more efficiently is valuable. Research has found that 94 per cent of employees encounter friction while preparing for, taking part in or following up on meetings. It also found that summaries can save employees an average of 1.6 hours each week by reducing time spent capturing actions, writing recaps and searching for information.
For a small business, that time matters. But a summary should be the beginning rather than the end. The bigger opportunity is to move from conversation to completion, with AI identifying what needs to happen next and helping carry the work forward.
It could turn an agreed action into a task, find information for a follow-up or pass an issue to the right person with the relevant context. The employee remains in control but no longer has to rebuild that context at every step.
What this looks like in practice
At Melbourne-based customer-service provider CDM Direct, agents handle complex calls covering everything from insurance claims to roadside assistance. These conversations require empathy and attention, but agents also need to collect customer information, find relevant guidance and complete admin work after each call.
By using AI to capture caller details, surface relevant information during conversations and automate call summaries, CDM Direct says it has reduced initial data-collection time from 120 seconds to as little as 10–20 seconds. Together with other workflow improvements, this has helped reduce its average call-handling time by around 30 per cent.
That is where AI’s productivity potential becomes tangible. The value is not simply having a transcript once the customer hangs up. It is helping employees find the right information, respond more efficiently and reduce the manual work that would otherwise follow every conversation.
Start with one conversation
A new parliamentary inquiry is examining what is preventing Australian businesses from adopting AI more deeply. Skills, investment and regulation will all be part of that discussion, but businesses don't need to wait for every answer or transform everything at once.
They can start with one conversation that happens regularly and look at the work it creates. How many steps sit between a customer request and its resolution? Which actions are agreed but later missed, and how much time is spent turning conversations into emails, tasks and updates?
From there, the business can choose one outcome that matters, whether it is a faster customer response, fewer missed actions or less time spent on administration, and assess whether AI is actually helping.
The goal is not to automate human conversation. Conversations are where relationships are built, ideas are tested and judgement is exercised. The opportunity is to make sure their value doesn’t disappear when the call or meeting ends.
Australia’s next productivity breakthrough won’t come from generating more summaries, emails or documents. It will come from using the knowledge already contained in our conversations to help people efficiently finish the work they started.