Will Accounting Be Automated?

Very likely, and it is already happening in pieces rather than all at once. We operate on a simple, firmly held assumption: almost 100% of accounting can be automated right now. All the technology is there, it just requires investment and time.

That distinction matters. The technology being ready is not the same as a firm being ready to use it. A single invoice carries somewhere between 80 and 100 small human decisions embedded in it, which vendor, which GL account, which tax treatment, whether the amount looks right. You cannot paste that logic into a general chatbot and get a usable answer. The hard part was never the software. It’s the accounting logic underneath it, and that only gets built by people who understand both the ledger and the code.

What’s already automated in a well run finance function today:
– Invoice capture and coding for standard, recurring vendors
– Bank feed matching and routine reconciliations
– Recurring journal entries and accruals
– Sample pulls for CRA or audit requests
– Repetitive reporting, like member or investor statement generation

What still needs a person: new vendor approval, unusual transactions, anything that smells like fraud, and every judgment call about what a number means for the client’s decisions.

Not the role. The job description underneath it, yes, is changing. We think of the future accountant less as someone entering data and more as a human orchestrator, someone who directs a set of digital tools and reviews their output, the way a senior manager today reviews a junior’s work. The value shifts from doing the task to being accountable for whether it was done correctly.

There is a counterintuitive wrinkle here. The common assumption is that AI takes junior jobs first. In practice, senior review work is often lower context and higher judgment, which is exactly where large language models are strong. Junior prep work involves ingesting messy, high volume source documents, which is where the technology still struggles. That means the skills worth building now are review, exception spotting, and client judgment, not faster typing.

You cannot put something into ChatGPT and get a perfect answer. It has to be built up, deliberately, by people who understand both the ledger and the code.

We built our own accounts payable automation internally over roughly six months, entirely by accountants who also code, not software engineers working from a spec. It now runs below human error rates across our client base. The lesson that mattered most wasn’t the accuracy. It was adoption: staff kept entering invoices by hand even after the tool was proven better, until we built a monitor that flagged manual entries and asked why. Technology readiness and human adoption are two different problems, and the second one is usually harder.

No, and this is where accounting behaves differently than most industries facing automation. In marketing, cheaper output tends to create more demand for marketing. In accounting, nobody wants more accounting. It’s a compliance function with a fixed scope: the books need to close, the return needs to be filed. So automation here mostly compresses cost and time rather than expanding the market. That’s a real shift in the profession, but it isn’t the same as the work disappearing. Someone still has to own the outcome, defend a position to CRA, and tell a client what their numbers mean.

We won’t put a number on anyone’s income, that depends on ownership structure, client base, and specialization, none of which we can estimate for you. What we can say is that the ceiling is rising for accountants who move toward judgment, oversight, and advisory work, and shrinking for anyone whose entire value is manual data entry. The earning path in this profession increasingly rewards people who can direct automated systems and explain results to a CFO or a board, not people who can process transactions faster than the next person.