A practitioner rang me recently with a question worth thinking about: were the charge rates in our benchmarking data starting to reflect the efficiencies firms are getting from AI? At this stage, they are not. Adoption across the profession is still mixed and, from what we are seeing, fairly conservative. But there is a more important issue sitting underneath the question, and it is not really about AI at all. It is about what happens to a fee when the work starts taking less time.

For accounting practices, this is not really an AI question. It is a pricing question.

If a job has always taken four hours and better systems reduce it to two, a firm billing strictly from time recorded will simply charge less for the job. Nobody has to approve a discount. Nobody has to write a policy. It happens at billing time, one job at a time, and often the first sign of it is a set of annual numbers that come in lower than they should.

I have watched a firm learn that lesson.

About twenty years ago a set of very good tax return and compliance templates came onto the market. A number of firms adopted them and, quite properly, became much more efficient.

One firm I knew went from approximately $1.1 million in annual fees to around $900,000 within twelve months. The partners initially struggled to understand what had happened. They had not lost a material number of clients, nothing significant had changed in the practice, and they certainly did not feel any less busy.

The explanation became fairly obvious once they looked at it. The work was taking less time and they were still billing largely from the time recorded, so the efficiency they had created was being passed directly to clients as a lower fee. The practice had invested in better systems and improved the way the work was done, and almost the entire financial benefit went elsewhere.

That story is not really about templates. It is about the relationship between time and price. If the fee is determined principally by how long the work takes, then reducing the time reduces the fee as well. It is an unusual commercial position to be in: becoming better at the work does not necessarily improve the economics of the job.

AI has the potential to create the same issue across a much broader range of work, and it is arriving at a point when many practices are already dealing with margin pressure, staff shortages and rising costs. If a set of templates cost that firm $200,000 in a year, it is worth asking what a broader efficiency shift could do to a practice that has not thought about it.

High-volume, relatively standardised compliance work would seem to be the most exposed, simply because repetitive processes are where these tools currently produce the clearest gains. If that is where the efficiency arrives first, firms need to know what they intend to do with it.

Leaving the answer to the timesheet is still a decision. It is simply an accidental one.

There is a second assumption sitting underneath this conversation, and it deserves some caution before anybody reprices anything.

AI does not reliably mean a job is suddenly completed in half the time. In most firms today, someone still has to prepare the input, review what comes back, check the work, correct errors and exercise professional judgement. That is real professional time, and it is often skilled time. What we are seeing first is frequently not a substantial reduction in hours, but an improvement in quality, consistency, or the ability to work through an issue more thoroughly.

Commercially, those are not the same thing. If a process genuinely becomes faster, you have an efficiency question. If the same amount of professional time produces a better outcome, you have a value question.

There are certainly more advanced systems being developed where a process can operate with considerably less human involvement. From what we are seeing, relatively few Australian practices appear to be operating at that level across large parts of their workflow yet.

So I would be careful about assuming a saving exists before you can measure it. You may well be producing better work. That does not automatically mean the job is costing you less.

There is also a straightforward commercial question here. The practice is paying for the technology. It is paying people to understand it, configure it, test it, review the outputs and manage the risks around it.

And this will not be a one-off investment. The systems firms use today will change. Processes will need to be reviewed. Staff will need to be trained. Somebody in the practice will need to remain responsible for making sure the system is still working properly. Those costs belong to the practice.

So if the investment eventually produces a genuine efficiency gain, I do not think there is any automatic reason the entire benefit should be handed to the client.

That does not mean clients should never share in efficiencies. There may be very good commercial reasons to pass some of the benefit through pricing, particularly in highly competitive work. But that should be a deliberate pricing decision, not an outcome your billing method produces on your behalf.

The client is ultimately buying an outcome: accurate work, sound advice, reduced risk and confidence that the matter has been handled properly. If you can deliver that outcome more efficiently while maintaining or improving the quality, you have created value. You have not created an obligation to reduce the fee.

For me, this is where the conversation becomes more interesting. The strongest argument for adopting these tools was never that firms could reduce headcount. Practices have spent years telling us they do not have enough capacity. So if technology genuinely gives some of it back, the better question is what the practice does with it.

There is usually no shortage of candidates. Advisory work that has been difficult to service. Clients who hear from the firm once a year and would benefit from a more regular conversation. Planning work that never quite gets done. Business development. Training and developing the people you hope will eventually become your managers and partners.

That has a direct bearing on practice value as well. When we look at practices, the quality of the fee base and the nature of the work matter as much as the top line. A firm that quietly allows its compliance fees to reduce year after year reduces its earnings, and in time what the practice is worth. A firm that maintains good fee quality and uses its additional capacity to deepen client relationships or grow advisory work is in a very different position. Most owners do not see that difference clearly until they come to sell.

None of this requires you to have a firm view about where AI is heading. It does require you to have a view about what your fee actually represents. If the answer remains simply the number of hours it took, then improvements in technology will begin making pricing decisions for you, and you will read about them in next year's accounts.