AI and How to Embrace the Human Element of Your Business

Here’s an opinion we often hear in AI conversations:

The ideal future is one where human involvement is minimal.

This means fully automated and hands-free. The accountant just reviews the output and clicks approve.

Want to know what we think about this?

We don’t buy it.

Not because the technology isn’t impressive, but because human judgement isn’t a gap in waiting to be filled.

It’s the whole point of the system.

The practices getting the most from AI right now aren’t the ones that have removed people from the process. They’re the ones that have been deliberate about where people belong in the process, and have built AI tools that make human review faster, sharper, and better informed.

Why are accounting firms investing in AI right now?

Admin is relentless in the profession and something has to give. Progress Software’s State of Accounting 2026 report surveyed over 300 accounting professionals and found that 75% say their workflows involve too many steps, and 65% report a lack of automation for routine tasks. That’s three-quarters of the profession spending more time than they’d like on work that doesn’t need a qualified accountant to do it.

AI fixes that. Document summarisation, deadline tracking, and client onboarding are tasks where speed and accuracy matter more than human judgement. Hand them to the right tools and you buy back hours that could spend on the work that grows your practice.

What does AI actually do well in an accounting practice?

AI genuinely delivers in a practice setting in the following areas:

• Processing and categorising large volumes of transactions

• Flagging anomalies, errors, or missing information before you file

• Drafting routine client communications and reminders

• Summarising documents, contracts, or financial statements

• Tracking deadlines across a large client portfolio

• Routing workflow tasks to the right person at the right time

AI is genuinely useful in accounting. A 2025 MIT/Stanford study found that accountants using AI could shift 8.5% of their time away from routine data entry towards higher-value work, and cut seven and a half days off their monthly financial close. That’s real time back. And according to Karbon’s State of AI in Accounting Report, 81% of accountants say AI has boosted their productivity.

But here’s the flip side. The same research shows that most of that productivity gain is coming from well-defined, structured tasks. The moment you move into territory that requires professional judgement, contextual knowledge, or ethical reasoning, AI’s performance drops sharply.

Where does the human element still win, every time?

The human element in your practice shows up in a few specific places:

• Judgement calls: AI can flag a discrepancy, but it can’t weigh up whether a client’s aggressive tax position is worth the risk given what you know about their circumstances.

• Client relationships: The accountant who calls a client proactively when something looks off, remembers their daughter just started a business, and gives a straight answer rather than a hedged one isn’t something you can automate.

• Advisory conversations: This is where the real growth is. The firms winning the advisory game are the ones who’ve used automation to create capacity for those conversations, rather than waiting until they have more time.

Accounting is a regulated profession with real personal liability attached to it. When a client faces a HMRC enquiry, the question isn’t which software made the decision. It’s which qualified professional determined this, and on what basis with what evidence.

‘AI calculated it’ isn’t a professional defence that would satisfy a PI insurer or HMRC. And it definitely won’t satisfy a client who’s just received a penalty based on incorrect advice.

An ICAS study on generative AI and professional judgement found that 72% of accounting professionals are worried AI could produce errors or incorrect decisions, and 64% believe auditors should validate AI outputs every single time they’re used to inform a professional conclusion.

The profession knows this, and the tools that are used need to reflect it.

Research consistently shows that 91% of people prefer a human to a chatbot when they need help with something that matters to them. A client who’s anxious about an HMRC enquiry, uncertain about a major business decision, or working through a tricky personal financial situation doesn’t need an automated response. They need an accountant who gets their situation, asks the right questions, and gives them advice they can act on. No AI can replicate these kinds of conversations.

How do you get the balance right between AI and the human element?

Start with a simple question for each task in your practice:

Does this require a trained professional’s judgement, or does it require accuracy and speed?

If the answer is accuracy and speed, automate it. If it’s judgement, keep a human in the loop.

In practice, the best approach is layered. Automate the data gathering, the first-pass processing, the reminders, and the routine filings. Then have your team review, advise, and build relationships on top of that cleaner, faster foundation.

What are the risks of leaning too hard into AI?

Two main ones, and both are worth taking seriously.

The first is over-automation. Automated reminders are fine. Automated responses to a client who’s just had a difficult year and is worried about their accounts is not fine. Set boundaries around what gets automated and what always gets a human response.

The second is skills atrophy. If junior staff never do the foundational work because AI does it for them, they don’t build the professional judgement that makes a great accountant. Use automation to remove drudgery, not to shortcut development. The goal is a team that’s better at the high-value work, not one that’s dependent on the tools to do the basic work for them.

How can you build an AI-positive culture without losing the human element?

In practice, that looks like this:

• Set clear boundaries: Which tasks AI handles, which require human judgement, and what review looks like at each handoff. Define it deliberately, don’t let it drift.

• Build your team’s confidence in interrogating AI output: The ability to review, question, and correct AI recommendations is as important as knowing how to use the tool in the first place.

• Be open with clients about how AI is used in their work: What safeguards exist, and how their data is protected. Most clients find this reassuring, not alarming.

• Measure the outcomes that matter: not how much has been automated, but how much time has been freed up for the work only your team can do.

AI is genuinely powerful, but it’s most powerful when it’s built around the expertise of the people using it. The human element isn’t a constraint on what AI can do. It’s what makes deploying AI a responsible, defensible, and ultimately more effective choice.

This content may have been developed with the assistance of AI tools. Content is reviewed by a member of the Bright team before publication. Guidance on legislation can be obtained from your industry body.