Virtual CFO work is one of the highest-value services an Australian accounting firm can offer. It is also one of the hardest to scale. Each engagement leans on a senior person to read the numbers, spot the risks, and turn them into decisions a business owner can act on. Add more clients and you add more senior hours, which is exactly the resource a growing firm has least of.
Claude changes the maths behind that constraint. Used well, it takes on the preparation and first-draft analysis that fills a virtual CFO's week, so the adviser spends their time on judgement and the client conversation rather than assembling the pack. This guide covers where Claude fits, what it should never touch, and the capacity numbers that make the case.
What a virtual CFO engagement actually involves
Before deciding where software helps, it is worth being honest about where the hours go. A typical monthly virtual CFO engagement for a business turning over $3M to $15M includes a recurring set of tasks:
Pulling the month's numbers from Xero or MYOB and checking they reconcile before anyone relies on them.
Building the management report: profit and loss against budget, balance sheet movements, and a short cash position summary.
Rolling a 13-week cash flow forecast and updating it for known payments, the next BAS, and any large receivables.
Writing the commentary that explains what moved and why, in plain language the owner understands.
Preparing three or four decisions for the board or owner meeting, each with the numbers behind it.
Of those, only the last two carry the real value a client pays for. The rest is preparation. On a $3,500 per month engagement, a firm might spend eight to twelve senior hours getting to the point where the adviser can actually think. That ratio is why most firms cap the number of virtual CFO clients a single adviser can hold at six or seven.
Where Claude fits in the workflow
Claude is strong at the preparation layer: reading structured and unstructured financial information, drafting clear commentary, and turning a pile of figures into a first version of the story. The pattern that works is to let Claude produce the draft and the adviser edit it, never the reverse.
Concrete tasks a firm can hand to Claude today include:
Drafting month-end commentary from the trial balance and prior-month report, flagging the three or four variances that matter most.
Turning a 13-week cash flow model into a plain-English narrative and a set of questions for the client.
Summarising a 40-page board pack into a one-page brief before the adviser reads the detail.
Preparing first-draft answers to the owner's likely questions, so the adviser walks in ready rather than reactive.
Building the first version of a budget or reforecast from last year's actuals and the client's growth assumptions.
None of these send Claude off to make a call on its own. Each produces a draft that lands on the adviser's desk in minutes instead of hours, which is where the capacity gain comes from.
The capacity maths
Here is the case in numbers. Say an adviser holds seven virtual CFO clients at an average of $3,500 per month, so $24,500 in monthly recurring revenue. Each client consumes roughly ten senior hours a month, which puts the adviser at about seventy hours before any new business development.
Move the preparation work to Claude and the per-client senior time drops to around six hours: the reconciliation review, the commentary edit, and the client meeting itself. The same adviser can now hold eleven clients in the same hours, lifting recurring revenue from $24,500 to about $38,500 a month, or an extra $168,000 a year, with no new hire.
Even if you are more conservative and assume the tooling only recovers two hours per client, that still frees enough senior time to take on two more engagements, worth roughly $84,000 a year. The investment to get there is a Claude subscription and a few days of building prompts and templates, not a $110,000 analyst salary plus on-costs.
What stays human
The reason this works is discipline about the boundary. Claude drafts; the adviser owns. Some things should never be delegated:
The final numbers. A person confirms the figures reconcile and the report is right before it reaches a client.
The advice itself. Recommending that a client delay a hire, chase a debtor, or change pricing is professional judgement, and under ASIC and Tax Practitioners Board expectations it needs a named human behind it.
Anything client-identifying your firm has not cleared for an external tool. Check your engagement terms and your obligations under the Privacy Act before uploading client data anywhere.
The relationship. The monthly conversation where an owner works through a hard decision is the product, not the pack that supports it.
Firms that hold this line get the capacity gain without the risk. Firms that let the draft quietly become the final answer are the ones that end up with a problem.
Getting started without betting the firm
The low-risk way in is to run Claude alongside your current process for one or two clients for a month. Have it draft the commentary and the board brief, then compare its output against what your adviser would have written. You will quickly see which tasks it handles well and which still need a heavy edit.
Most Australian firms we work with in Sydney and beyond find the month-end commentary and the board-pack summary are the fastest wins, because they are high-effort, low-judgement, and repeat every single month. Prove the time saving there first, then widen it to the forecasting and budgeting work.
If you want help setting this up for your firm, book a short call and we will map your virtual CFO workflow to a Claude setup you can trust.
