The accounting profession is not immune to the AI wave, and it shouldn't try to be. The honest reality is that a significant portion of what accountants have traditionally charged for — data entry, transaction categorisation, report generation, basic compliance work — is being automated at a pace that will only accelerate. That's not a threat to good accountants. It's a reallocation of time from low-value mechanical tasks to high-value thinking.
For business owners, the question isn't whether AI changes accounting. It obviously does. The questions worth asking are: what does it actually change, what stays the same, and how do you tell whether your accountant is using it well?
What AI genuinely does well in accounting
The areas where AI is having real, measurable impact in accounting practices right now are worth understanding specifically rather than in vague terms.
Transaction processing and categorisation
This is the most mature application. Cloud accounting platforms — Xero, QuickBooks, and others — have been using machine learning to categorise bank transactions for years, and the accuracy has improved substantially. What used to require a bookkeeper manually coding hundreds of lines each month now happens largely automatically, with human review focusing on exceptions. For businesses with high transaction volumes, this is a genuine time and cost saving.
Document processing and data extraction
Receipt capture, invoice processing, and document OCR have reached a level where a photo of a receipt taken on a phone can be automatically matched to the relevant supplier, coded to the right expense category, and reconciled to the bank statement with minimal human involvement. Again, this is not a future capability — it's available now in the major accounting platforms.
Anomaly detection and error-flagging
AI systems are considerably better than humans at spotting patterns that don't fit — duplicate invoices, transactions that fall outside normal ranges, supplier payments that look unusual. This kind of automated checking provides a layer of oversight that would be prohibitively expensive to do manually at scale.
Financial Q&A and plain-English reporting
The more recent development — and the one Insights AI is built around — is the ability to connect AI to live financial data and allow business owners to ask questions in plain English. "What was my best month last year?" "Who are my top five clients by revenue?" "Am I on track against budget?" These questions used to require either a finance team or a time-consuming manual analysis. A well-built AI layer answers them instantly, from the same data that sits in the accounting system.
The genuine value of AI in this context isn't replacing the accountant — it's giving business owners visibility into their own numbers that they previously couldn't access without asking. That changes the quality of decisions made between accountant conversations.
What AI cannot do — and won't any time soon
The limitations are as important to understand as the capabilities, because they define what still requires human expertise.
Judgement on complex or ambiguous situations
Tax law is not a series of clear rules that AI can reliably apply. It's a body of legislation, case law, HMRC guidance, and practice that contains ambiguity, grey areas, and situations where reasonable professionals disagree. When a business owner faces a non-standard situation — a novel transaction structure, an unusual asset, an HMRC enquiry — the right answer requires judgement built on experience. AI can surface relevant information. It cannot make the call.
Strategic financial advice
Deciding whether to incorporate, how to structure a management buyout, whether to take on debt or equity financing, how to plan for an exit — these decisions require understanding the numbers, the business, the owner's personal circumstances, and their goals. That understanding is built through a relationship, not a data feed. AI assists here but doesn't replace the conversation.
Relationship and trust
Good accountants know things about their clients that aren't in the accounts. They know when the business is under stress, when the owner is making decisions from fear rather than strategy, when something in the numbers doesn't match what's being said. That knowledge comes from paying attention over time. It's not something that can be automated.
Accountability
If an AI system gives incorrect tax advice and you act on it, the consequences fall on you. A qualified accountant carries professional liability insurance, is regulated by a professional body, and can be held accountable. AI tools, however capable, operate without this framework.
What it means for business owners choosing an accountant
If AI is automating the mechanical parts of accounting, the cost of delivering basic compliance services should fall. Accountants who are not using AI tools to improve efficiency will find it increasingly difficult to justify the same fee levels as those who are. That's good for clients.
At the same time, what you're really paying for — and what should command a premium — is the expertise, judgement, and relationship that AI cannot replicate. An accountant who has used AI to eliminate the drudge work and redirect that time to proactive advice is more valuable than one who is just doing the same mechanical tasks more slowly.
Some practical questions worth asking your accountant, or a prospective accountant:
- What tools are you using to automate bookkeeping and transaction processing?
- Can I get access to my financial data in a way that lets me ask questions without waiting for a monthly report?
- How has AI changed the way you work, and has any of that efficiency been passed back to clients?
- What do you spend your time on now that software handles the mechanical work?
The answers — and the quality of thinking behind them — will tell you more about the quality of the accountant than any accreditation.
Where Insights fits in
We built Insights AI because we believed business owners deserved access to their own financial data without having to ask for it. The AI layer connects directly to the client's live accounting data and answers questions in plain English — available around the clock, not just when we're in the office.
That doesn't replace the monthly conversations about strategy, the tax planning reviews before 5 April, or the call when something unusual is happening in the business. It supplements them. The goal is that clients at Insights have better visibility between our conversations, make better-informed decisions, and come to those conversations better prepared.
AI is changing accounting. But the parts of accounting that matter most — the thinking, the advice, the relationship — remain stubbornly human. That's not a limitation. It's the point.
