AI in Accounting Firms: Why the Client Relationship Still Wins
Used as a shortcut, it quietly erodes the one thing clients pay you for. Used well, it is the best chance in a generation to be the firm they will not leave.
Every few years this profession finds a new answer to the same question: how do we do more with less? Fifteen years ago the answer was offshoring. Today it is artificial intelligence. I have watched enough of these waves to know that the tool is rarely the problem — what matters is whether you treat it as a shortcut or as a way to do the part of the job only you can do, better.
- AI is the newest version of an old temptation in accounting — a cheaper, faster way to do the work, adopted before anyone has costed the consequences.
- A firm that uses AI to remove its people from client work is eroding the one asset it cannot rebuild: trust.
- AI produces information; it does not carry the duty of care. The professional who signs the advice still owns every word of it.
- The next five years are the best opportunity in a generation to deepen client relationships, and AI used well buys back the time to do exactly that.
- The most valuable firms in five years will be the ones using AI to be more present with clients, not the ones using it to disappear.
I am not against artificial intelligence. Used properly, it is the most useful efficiency tool to arrive in this profession in a long time. My concern is not the technology. My concern is the pattern — because I have seen this profession reach for a shiny object before, and I have watched what it cost the firms that grabbed it without thinking.
When offshoring arrived, the promise was the same: do the work cheaper, free your people up, protect your margin. The firms that did it well built their own dedicated teams and stayed close to the work. The firms that bolted it on as a pure cost play often found that the savings came with a hidden bill, paid years later. AI is following the same curve, only faster.
AI Is the Newest Answer to an Old Temptation
The appeal is obvious. Compliance work is repetitive, margins are tight, and staff are hard to find. A tool that does the dull, repeatable work in a fraction of the time looks like a gift. For a great deal of that work, it genuinely is.
But there is a quieter consequence that very few firms costed into the decision, and it is the same one offshoring created. A decade of sending compliance work offshore quietly hollowed out the supervisor-to-manager tier — the layer that holds client relationships day to day and becomes the next generation of partners. And it became self-perpetuating: a firm cannot stop outsourcing because it no longer has the resources to do the work itself, yet outsourcing is part of why it never developed those people in the first place. Chicken and egg — a hamster wheel the profession has been running on for years. The short-term fix quietly costs the long-term profession.
AI is that same temptation arriving one rung lower. The repetitive, rules-based work it does so well is exactly the work graduates have always learned on — and you cannot hand a graduate an advisory role. They earn it on the simple work first. Strip that work out and they miss the learning and never build the judgement to step up. I deal with the broader workforce picture in Do You Have a Succession Plan? — here it is enough to say the work you automate away is also the work your people learn on. That is the cost side of the ledger, and the opportunity side is more interesting.
We have a recent precedent for what happens when a profession reacts to pressure and pays for it later. After the post-Royal-Commission reforms, the number of licensed financial advisers in Australia fell from a peak of around 28,000 in 2018 to roughly 15,000 — almost halved in under a decade — and the pipeline of new entrants is nowhere near replacement levels. That was a regulatory shock rather than a technological one, but the lesson carries: a short-term response to pressure can hollow out a profession for a generation.
The tool is never the strategy. How you use it is.
AI Does Not Remove the Duty of Care
In 2025, Deloitte agreed to repay part of a roughly $440,000 fee to the Australian government after a report it delivered was found to contain fabricated references, citations to research that did not exist, and a quote attributed to a federal court judge that was never said. The firm later disclosed that generative AI had been used to help write it. This is one of the largest names in the profession, and it still went out the door.
I see the smaller version of this in practices all the time. A junior is asked to prepare tax planning for a client, runs it through an AI tool, and brings back recommendations that are confidently wrong. The real failure is not that the machine made things up — they do, and that is now well understood. The failure is that no one checked. When a partner signs off on advice, the duty of care sits with the partner, not the software. A client who is poorly advised will not accept "the AI told us so" as an answer, and neither will a court or a professional body.
There is a second risk that worries me as much as accuracy, and it gets far less attention: where does the information go? When firm and client data is fed into a tool, who can see it, who controls it, and is there any security around it at all? For a profession that holds some of the most sensitive financial information its clients own, that is not a question you can leave unanswered.
If your firm is using AI but has no written policy on who may use it, what must be checked before it leaves the building, and what client data may never be entered into it, you do not have an efficiency gain. You have an unmanaged risk that happens to be fast.
The Relationship Is the Moat
Here is the part that should make every principal optimistic rather than anxious. AI can assemble information beautifully. It cannot sit across the table from a worried client at the end of a hard financial year and tell them, with judgement and authority, what it actually means for them. People do not want a computer to interpret their life. They want a person they trust.
We all know the experience of someone arriving at the doctor having searched their symptoms online, convinced they have everything from the trivial to the terminal. The information was free and instant. The reassurance, the judgement and the plan were not — and that is exactly the role a good accountant plays. The firms that understand this will use AI to clear the routine work off their desks so they can spend more time doing the thing a machine never will: knowing the client.
The shortcut firm
- Uses AI to cut hours and headcount, then pockets the saving.
- Pushes more client contact to email and portal, less to conversation.
- Treats AI output as the answer rather than a first draft.
- Wins on price until a competitor wins on relationship.
The relationship firm
- Uses AI to clear routine work, then reinvests the time in clients.
- Adds touchpoints — the call, the review, the proactive heads-up.
- Treats AI output as a draft a qualified person always checks.
- Becomes the firm clients will not leave, and buyers compete for.
No one will ever trust a computer to give them advice. They will trust the person who used the time the computer saved to call them first.
Use AI to Buy Back Time for Clients
The practical move is not complicated, though it does take discipline. Decide deliberately where AI saves you time, and then decide just as deliberately where that time goes. If it disappears into a slightly higher margin and nothing else changes for the client, you have taken the shortcut. If it goes back into client relationships — the proactive call, the half-hour debrief, the advice offered before it is asked for — you have built something a competitor cannot easily copy.
What does that look like in practice? It might be letting AI take the first pass at routine, non-client work — drafting a standard letter or template, or turning a dense piece of new legislation into a plain-English summary you then tailor — so the hour you save goes into the client meeting rather than the admin. It might be using it to make your existing systems talk to each other and strip out the manual handling in between. The firm-changing uses are usually the unglamorous internal ones: done inside a tool you control, kept well clear of confidential client information, and checked by someone qualified before it counts.
Put a simple policy around it. Name who may use AI and for what. Require that a qualified person reviews anything before it reaches a client. Be explicit about what client data may never be entered into a public tool. And keep developing your people on the judgement-heavy work, because the value you sell when you eventually transition the practice is, in large part, the strength of those relationships and the team who hold them. A buyer will pay a premium for a firm whose clients are loyal to the firm, and discount one whose clients are loyal only to a departing principal — a point I have made in detail about what the market actually pays for.
This month, pick one thing AI now does faster in your firm and decide where the saved time will go. Put it into a client conversation that would not otherwise have happened — and write down the one rule everyone must follow before AI output leaves the building. That is a strategy. Everything else is just the tool.
AI in Australian accounting firms
Will AI replace accountants?
Not the good ones. AI is very capable at the repetitive, rules-based work, and that work will increasingly be automated. What it cannot do is carry professional judgement, take responsibility for advice, or hold a relationship with a client who wants to be understood, not processed. The accountants at risk are the ones who only ever did the work a machine can now do. The ones who advise, interpret and reassure are about to become more valuable, not less.
Should my accounting firm have an AI policy?
Yes, and it does not need to be long. At a minimum it should state who in the firm may use AI tools and for what, what must be checked by a qualified person before anything goes to a client, and what client information may never be entered into a public tool. A firm using AI without those three rules is carrying an unmanaged risk. The policy is what turns AI from a liability into an efficiency gain.
Is it safe to put client data into AI tools?
Not into a public, consumer-grade tool. Accounting firms hold some of the most sensitive financial information their clients own, and once data is entered into an open tool you often cannot say where it goes or who can see it. If AI is going to touch client data, it needs to be in an environment where confidentiality and control are genuinely understood and documented. When in doubt, keep client-identifying data out of it.
How do I use AI without damaging the client relationship?
Use it to remove work, not people. Let AI clear the routine tasks, then deliberately reinvest the time you save into client contact — the call, the review, the proactive advice. The firms that lose ground are the ones that use AI to push more of the relationship onto email and portals. The firms that win use the saved time to be more present, which is the one thing competitors and software both struggle to copy.
Does using AI affect my practice's value when I sell?
It can go either way. Sensible, well-governed use of AI that frees your people up for client work strengthens the business and its relationships, which supports value. Using AI to strip out the people and the contact that hold clients to the firm tends to weaken the very thing a buyer is paying for. Buyers look closely at whether clients are loyal to the firm or only to a departing principal, and the relationship is where that question is answered.
Decide where the time you save actually goes.
We have advised Australian accounting, financial planning and legal practices for over 50 years. If you want a clear-eyed view of how to use AI to strengthen your firm rather than hollow it out, let's talk.
Our Consulting Services Talk to Michelle
What practices are selling for, and what is moving the numbers
Michelle’s monthly note on practice valuations, buyer activity and the conditions we are seeing across accounting, financial planning and legal firms.
