Most conversations I have with business leaders about AI, suffer from the same problem. Everyone is talking about it, but nobody is talking about the same thing.

One person means the chatbot their marketing team is using to draft social posts. Another means the autonomous system their competitors are quietly deploying to handle customer triage. A third has just come back from a conference where someone made a convincing case that we’re five years from AI that makes humans economically redundant across whole industries. They’re all calling it “AI.” However they’re describing completely different worlds with significantly varied outcomes for business and society.

This matters because the decisions business leaders make today — about investment, headcount, capability building, strategy — depend heavily on which version of AI you think you’re navigating. Get that wrong and you’re either asleep at the wheel or burning resources preparing for a future that’s further away than you think.

So let me lay out three distinct scenarios. Not predictions — I’m not going to pretend anyone knows exactly how this unfolds. Albeit there are those that do. But useful frames for thinking about what your business should actually be doing.

Scenario 1. The AI Many Already Have: Powerful Tools, Human Hands on the Wheel

This is where we are. AI systems today are genuinely impressive at specific, bounded tasks — writing, summarising, generating code, analysing data, creating images. They’re fast, cheap, and getting better quickly. But they don’t think. They don’t set goals. They need a human to define the problem, review the output, and catch the inevitable errors.

Tools like Claude, Copilot, and dozens of specialised vertical products are already inside most organisations in some form — sometimes officially, often not. And the productivity gains for individual knowledge workers are real. Research puts it somewhere in the range of 20 to 40 percent improvement on tasks like drafting, research, and synthesis. That’s not nothing.

What I see, though, is a wide gap between organisations treating this as a productivity experiment and those genuinely redesigning how work gets done. The first group is getting incremental efficiency. The second is compressing timelines, reducing handoffs, and delivering faster to clients. The gap between them is already meaningful and it’s widening.

The risk at this stage isn’t really about technology. It’s about leadership complacency. “We’re watching the space carefully” has become the new “we’re exploring a digital strategy” — something leaders say when they haven’t actually committed to doing anything. The businesses that are building genuine AI fluency and redesigning workflows around it right now are accumulating an advantage that compounds. That process takes time, and the clock is running.

One thing worth noting: the winners at this level aren’t necessarily the biggest companies or the ones with the most data. A small professional services firm, NDIS provider or Real Estate Agent that genuinely embeds AI into how it delivers work can outmanoeuvre a much larger competitor that’s still treating AI as an IT project.

Scenario 2. The Next Frontier: When AI Starts Pursuing Goals, Not Just Completing Tasks

Something qualitatively different is starting to emerge, and most businesses aren’t ready for it.

For most, current AI use requires you to define a task, hand it over, check the result. The next generation — already in early deployment in software development, research, and some customer-facing environments — can be given a goal and figure out the steps itself. These “agentic” systems can use tools, make decisions, interact with other systems, and hand back to a human only when they hit a genuine wall, or when the human specifies. The difference between “write me a summary of this document” and “go and research our top three competitors, synthesise what you find, and flag anything that changes our pricing assumptions” is enormous.

Let’s take it a step further to a framework in use by Quigly AI: “You are my product pricing expert, deploy three sub-agents to run independently to go and research our top ten competitors, synthesise what they find, and flag anything that changes our pricing assumptions. Then ‘you’, as my pricing expert will collate this data, assess the findings and write me a summary document including recommendations.”

When AI can pursue goals rather than just complete binary tasks, the economics of entire business functions shift. A small team overseeing a set of AI agents can do work that previously required a much larger headcount. That’s not theoretical — it’s in use right now.

This is where business models start to crack. Not just processes — models. The professional services firm that charges by the hour for work that an AI system can do in minutes has a fundamental problem. The large enterprise with layer upon layer of coordinators and information-relay managers has a structural problem. These aren’t comfortable observations, but they’re honest ones.

Middle management — as a coordination and information-relay function — becomes increasingly hard to justify. What doesn’t diminish in value is judgement, accountability, and the ability to set meaningful direction. The leaders who understand this distinction will make very different decisions about their organisations than those who don’t.

There’s also a governance dimension here that’s moving faster than most boards appreciate. When an AI system is making decisions and taking actions on your behalf — interacting with customers, executing on data, initiating processes — you need to know where the accountability sits, what the audit trail looks like, and what the failure modes are. These aren’t IT questions. They’re board questions.

Another big consideration is who will our competitors be six months from now? The new era of AI brought with it the prediction that the world’s first billion-dollar one-man company would come to fruition, and that has occurred. One man and his AI army can massively disrupt legacy industries.

Scenario 3. AGI: The Scenario Most Don’t Want to Think About and Everybody Should

AGI — Artificial General Intelligence — is the scenario where AI can reason, learn, and apply knowledge across any domain the way a capable human can. Matching or exceeding human cognition across virtually all domains. Not specialised in one area. Not dependent on how it was trained for a specific task. Genuinely general.

Nobody knows when this arrives, or whether it arrives at all in the form people describe. The serious researchers are split across a wide range — some say a decade. In a podcast that I listened to yesterday, with some seriously intelligent people on the subject, several of them were saying we are looking at a two-year horizon. And even then the definition itself is contested.

But here’s why it deserves serious thought even with that uncertainty: the decisions that prepare a business for an AGI world are largely the same decisions that serve you well in the agentic world. So thinking through the implications costs you very little, and it sharpens your strategic thinking considerably.

In an AGI scenario, the scarcity logic that underpins most professional knowledge businesses breaks down. If a system can acquire expertise in any domain instantly and apply it without error, the premium on human specialist knowledge shrinks dramatically. That affects law, accounting, consulting, medicine, financial advisory — any field built partly on the fact that expertise is rare and takes years to develop.

Innovation cycles would accelerate in ways that are hard to internalise. We’re already seeing AI compress research and development timelines significantly. At AGI level, that compression becomes extreme. First-mover advantages either become permanent — because you’ve embedded the capability before anyone else — or collapse entirely, because a competitor can close a five-year gap in weeks or days.

The access question is one I think about a lot. The most significant business risk in an AGI world may not come from competitors in your industry. It may come from the concentration of AI capability in the hands of a very small number of companies that control the infrastructure. How that access gets priced, and how it gets regulated, will determine whether AGI becomes a broadly available business input or a structural advantage locked up by the few who control it. That’s a political and regulatory question as much as a technological one, and business leaders who engage in those conversations now will have far more influence over the outcome than those who show up late.

What doesn’t become obsolete in an AGI world is trust. At least not initially. Relationships built over time. The ability to make ethical calls in ambiguous situations. Cultural intelligence. Creativity grounded in genuine human experience. These things are hard to replicate precisely because they’re not reducible to information processing. The organisations that have invested in these dimensions will find they have a different kind of advantage — less obvious, but more durable.

So What Do You Actually Do?

I’m wary of wrapping this up with a tidy action list, because the honest answer is that no one has a clean playbook for navigating this. But a few things seem clear regardless of which scenario plays out fastest.

The businesses that will fare best are those that build genuine AI literacy across the organisation — not just in the technology team, but at board level, in the executive group, and in the people who run client relationships and operational functions. Not so everyone becomes a technologist, but so that the people making consequential decisions understand what they’re deciding about.

The organisations that get the augmentation framing right will outperform those that think of AI primarily as a replacement technology. Humans and AI working together — with AI handling volume, speed, and pattern recognition, and humans handling judgement, accountability, and relationships — is a more powerful combination than either alone. The companies that figure out the right division of labour will pull ahead.

And start the governance work. Not because regulators are forcing you to, but because the organisations that have clear frameworks for how AI is used, how decisions get reviewed, and where accountability sits will be more resilient as the technology advances. It’s also increasingly a factor in how clients and talent assess you.

AI strategy isn’t a subset of IT strategy anymore. If it’s sitting in your technology function as a capability initiative, it’s in the wrong place. This is a business strategy question, and it belongs at that level.

If people are frightened by this picture I think that’s reasonable. However, the genie is out of the bottle and it’s not going back in. So I think the leaders who engage with it seriously — who sit with the uncertainty rather than waiting for a clearer answer — will make better decisions for their organisations than those who treat it as someone else’s problem or wait for greater clarity.

The shape of what’s coming is clear enough. The specifics are not. Learning to operate well in that gap is probably the most valuable capability a business leader can build right now.

There’s a gap in most accounting and bookkeeping practices right now — and it’s not the one you might expect.

It’s not a skills gap. It’s not a technology gap. It’s a looking gap.

Most practice owners I speak with know that AI is doing something useful in businesses like theirs. They’ve read the articles. They’ve sat through the webinars. They have a vague sense that there’s probably something in it for their practice too.

But they haven’t looked. Not properly. Not at their own workflows, their own bottlenecks, their own specific situation.

That’s the gap. And in 2026, with AI adoption accelerating faster than most people expected twelve months ago, it’s starting to matter.

There are a variety of reasons why they are not looking. In some cases it will be because they don’t know what to look for, or in others they don’t know what to do with what they find, which is perfectly reasonable but also not an excuse for inaction.

This isn’t an article about the future of accounting. It’s about what’s happening right now — in practices of similar size to yours — and what a realistic, practical look at AI actually involves.

Where the Time Is Actually Going

Before talking about AI, it’s worth being honest about where practice time disappears.

In almost every accounting and bookkeeping firm I’ve worked with, the answer is the same. It’s not the complex work. It’s the repetitive work that accumulates invisibly until it’s consuming hours nobody can account for.

Here’s what it typically looks like:

Client onboarding. The same information collected the same way for every new client. Emails back and forth. Documents requested, chased, received in the wrong format, converted, filed. A process that should take thirty minutes routinely takes three to four hours — not because it’s difficult, but because it’s largely manual and nobody has ever had time to redesign it.

Reconciliation. Data arriving from multiple sources in multiple formats. Manual matching. Exceptions that interrupt the flow. The work itself isn’t hard — it’s just relentless, and it scales directly with client volume in a way that advisory work doesn’t.

Report generation. Templates rebuilt or reformatted for each client. Data pulled from one system, formatted in another, reviewed, adjusted, exported. The final product takes an hour to produce. The surrounding process takes three.

Document chasing. Follow-up emails. Reminder messages. Status tracking. The administrative overhead of getting clients to provide what you need, when you need it, in a format you can actually use.

Data movement between platforms. Information entered in one system, re-entered in another. Exports, imports, reformats. Work that exists only because the systems don’t talk to each other — and that someone has to do manually every time.

Communications. The bedrock of quality professional services; however, so many organisations do it poorly and, importantly, do it manually. As a result, important communications are simply never sent to clients, which at times can fracture relations.

None of these are complex problems. None of them require deep expertise to resolve. They persist not because they’re difficult to fix but because fixing them requires carving out time that busy periods never provide.

The practices that have started looking at AI aren’t finding exotic solutions. They’re finding that several of these workflows — the ones that feel like background noise — can be substantially automated with tools that already exist.

What AI Actually Does (and Doesn’t Do) in a Practice Environment

Let’s be direct about this, because the gap between AI hype and AI reality is where most practice owners get lost.

AI & Automation is not going to replace your accountants. Not yet anyway. It is not going to make complex advisory judgements. It is not going to manage client relationships or handle the work that requires genuine professional expertise.

What AI & Automation is genuinely good at — and what makes it immediately relevant to accounting and bookkeeping practices — is structured, repetitive, high-volume tasks. Exactly the category of work that consumes the most hours in a typical practice and delivers the least value per hour spent.

Document processing. Data extraction and formatting. Workflow routing. Automated follow-up sequences. Report assembly from structured inputs. Client communication templates triggered by specific events or milestones.

These are not glamorous applications. They are not the AI stories that make the news. But they are the applications that recover hours — real, measurable hours — in practice environments where capacity is the binding constraint.

The other thing worth saying clearly: the investment required is smaller than most practice owners expect. The AI adoption conversation often conjures images of large technology projects, significant consulting engagements, and months of disruption.

The reality, for most of the workflows described above, is a series of small, specific projects. Each one takes days or weeks, not months. Each one targets one workflow, fixes one bottleneck, and generates a measurable return in recovered time.

AI adoption for accounting firms in 2026 is not a transformation programme. It’s a series of practical decisions about which repetitive tasks are worth automating first.

What Early-Adopting Firms Are Actually Doing

The practices moving fastest right now are not the ones with the biggest technology budgets. They’re the ones that decided to look — and then acted on what they found.

Here’s what small, practical AI projects look like in accounting and bookkeeping environments:

Client onboarding automation. A bookkeeping firm recently reduced their onboarding process from four hours to under an hour. They didn’t buy a new platform. They mapped their existing onboarding steps, identified three that were entirely manual and entirely repetitive, and built automated workflows to handle them. Document requests triggered automatically. Follow-up reminders sent without anyone initiating them. Information captured in structured format and filed without manual data entry. The project took less than a week to implement.

Reconciliation assistance. Practices using AI-assisted reconciliation tools are processing the same volume in materially less time. The AI handles the routine matching — the straightforward transactions that make up the bulk of the queue. Exceptions are flagged for human review. The accountant’s time goes to the work that actually requires their judgement.

Automated reporting. Firms that have built templated report generation workflows are producing client reports in a fraction of the previous time. Data flows from source systems into structured templates. The accountant reviews, adjusts where necessary, and approves. The assembly work — which previously consumed hours — is largely gone.

In each of these cases, the financial impact follows a consistent pattern. Recovered hours per team member per week. Multiplied across a team of five or ten people. Across a full year. The number is rarely dramatic in any single week. Compounded over twelve months, it shows up meaningfully in margins, in capacity, and in what the practice can take on without adding headcount.

The practices that haven’t looked yet aren’t in crisis. But the gap between them and the ones that have is widening — quietly, consistently, one automated workflow at a time. How wide will that gap be in 12 months time? 24 months?

How to Find Your Own Opportunities

The starting point isn’t technology. It’s observation.

Most practices have three to five AI opportunities sitting in their current operations right now. Not hidden — just unexamined. The workflows that feel like background noise are usually the ones worth looking at first.

Three questions worth sitting with:

What does your team do every week that is entirely repetitive and entirely manual? Not the work that requires expertise or judgement — the work that follows the same steps every time, with the same inputs, producing the same outputs. If you could describe it as a checklist, it’s worth examining.

Where does work slow down or back up during busy periods? The bottlenecks that appear every tax time, every end of quarter, every reporting cycle — these are pointing directly at the workflows where automation would have the most immediate impact.

What would you look at if you had a week to focus on nothing else? Most practice owners have a mental list of things they’ve always meant to examine but never had time for. That list is usually a reasonable proxy for where the opportunities are.

Answering these questions honestly takes 1–2 hours. Acting on the answers — finding the right tools, scoping the right projects, sequencing them in the right order — is something you can probably figure out for yourself if you have the time and some technical knowledge, however for many this is where external expertise is genuinely useful.

An AI Opportunity Assessment is designed exactly for this moment. A focused 90-minute session working through your specific workflows. A written report within 48 hours covering where your opportunities are, what the financial impact looks like, what tools are relevant to your practice, and a prioritised AI adoption roadmap for acting on what we find. The cost is $1,500 excl. GST.

It’s not a large commitment. It’s a clear-eyed look at what’s actually available to your practice — without the hype, without the jargon, and without a lengthy consulting engagement attached to it.

The Honest Closing Thought

The accounting and bookkeeping firms that will look back on 2026 as a turning point aren’t doing anything exotic. They’re doing something simple: they’re looking at their operations honestly, finding the workflows that don’t need to be as manual as they are, and fixing them one at a time.

The technology is ready. The applications are practical. The investment is manageable.

The only thing most practices are waiting for is the decision to look.

If you’d like to understand what an AI Opportunity Assessment involves before committing, I’m happy to have a 20-minute discovery call first. No obligation — just a straightforward conversation about your practice and whether there’s something worth exploring.

You can reach me at team@quigly.com.au or find out more at quigly.com.au.

Mark Williams is the founder of Quigly AI, an AI consulting firm helping Australian SMBs discover and act on their AI opportunities. Based in Melbourne VIC, delivering nationally.

Seven weeks.

That’s what’s left before the end of the financial year — and for most accounting and bookkeeping practices, the next seven weeks will follow a predictable pattern. The workload builds. The team stretches. The manual processes that have always been a problem become impossible to ignore.

EOFY is the most revealing moment in the accounting calendar. Not because it introduces new problems, but because it makes existing ones visible in a way that quieter periods don’t. The reconciliation backlog, the reporting pressure, the document chasing, the client communications that didn’t happen — they’re not abstract right now. They’re sitting in your team’s inbox today.

That visibility is actually an opportunity. Not just to get through EOFY, but to come out the other side with a clearer picture of where your practice is leaking time — and what to do about it.

The EOFY Pressure Pattern

Every practice principal I’ve spoken with describes the same EOFY experience. It’s not that the work is harder than usual. It’s that the volume of repetitive, manual work compounds faster than the team can absorb it.

Here’s what that typically looks like in the weeks before 30 June:

Reporting. Clients who haven’t asked about their financials all year suddenly need everything at once. Management reports, BAS reconciliations, year-end summaries. Each one requires pulling data from multiple sources, formatting it consistently, and reviewing it before it goes out. The work itself isn’t complex — but there’s a lot of it, and it scales directly with your client base.

Document collection. The annual hunt for missing information. Bank statements, receipts, payroll records, signed authorities. Follow-up emails sent manually. Reminders written individually. The same requests going to the same clients who were slow last year and will be slow again next year.

Reconciliation backlogs. Transactions that have been sitting in the queue because quieter weeks never quite delivered the time to clear them. They accumulate. By late May, the backlog is a problem that has to be resolved before 30 June regardless of everything else competing for attention.

Client communications. Updates that should have gone out earlier in the year but didn’t — because there was always something more urgent. EOFY is when the cost of that silence shows up. Clients who haven’t heard from you in months are not in the best frame of mind to provide what you need quickly.

None of these are new problems. What EOFY does is compress them into a window where they all demand attention simultaneously.

What AI and Automation Actually Does in a Practice Environment

Before getting to what firms should do before 30 June, it’s worth being direct about what AI and automation actually means in a practice context — because the gap between the hype and the reality is where most practice owners get lost.

AI is not going to replace your accountants. It is not going to make complex advisory judgements or manage sensitive client relationships. What it is genuinely good at is structured, repetitive, high-volume work — exactly the category of tasks that consumes the most hours in a practice during EOFY and delivers the least value per hour spent.

Document requests triggered automatically rather than written manually. Follow-up reminders sent on schedule without anyone initiating them. Report templates populated from structured data sources rather than built from scratch each time. Client communications drafted and sent based on workflow milestones rather than when someone remembers.

These are not glamorous applications. They are the applications that recover hours — real, measurable hours — during the weeks when your team has the fewest hours to spare.

The investment required is also smaller than most practice owners expect. We are not talking about a technology overhaul or a six-month implementation project. The applications above are achievable with tools that already exist, configured to your specific workflows, in days or weeks rather than months.

What Firms Are Actually Doing Before 30 June

The practices moving into EOFY in a stronger position than previous years aren’t doing anything exotic. They’ve made a small number of specific, practical decisions about which manual processes are worth automating first.

Automated document collection. Rather than manually chasing clients for missing information, practices are using tools that trigger requests automatically, send reminders on a schedule, and collect documents through a structured intake process. The time saving per client is modest. Across a client base of fifty or a hundred, it’s material.

Report assembly workflows. Firms that have built templated reporting workflows are producing the same reports in a fraction of the time. Data flows from source systems into structured templates. The accountant reviews, adjusts where necessary, and approves. The assembly work — previously consuming hours — is largely automated.

Proactive client communication sequences. Rather than relying on individuals to remember to send updates, practices are running automated communication sequences triggered by workflow milestones. Clients hear from the practice at the right moments without anyone having to initiate it manually. The impact on client satisfaction and document turnaround is significant.

Reconciliation tools. AI-assisted reconciliation handles the routine matching — the straightforward transactions that make up the bulk of the queue. Exceptions are flagged for human review. The accountant’s time goes to the work that actually requires their judgement.

In each case, the financial impact follows a consistent pattern. Hours recovered per team member per week, across a team, across the seven weeks between now and 30 June. That translates directly into capacity — the ability to handle the EOFY volume without the team working the hours they worked last year.

What to Do in the Next Seven Weeks

The question isn’t whether AI and automation are useful for accounting and bookkeeping practices. The evidence is clear that they are. The question is what to prioritise in the time available before 30 June.

Here’s a practical framework:

Weeks 1–2: Map where the time is actually going. Before choosing any tools, spend two hours honestly mapping your team’s recurring workflows. Which tasks happen the same way every time? Where does work back up during peak periods? Which manual processes consume the most collective hours? This mapping exercise identifies your highest-value automation targets — and in most practices, three to five of them become immediately obvious.

Weeks 3–4: Prioritise by impact and effort. Not all automation opportunities are equal. Some take days to implement and recover hours every week. Others are technically interesting but practically marginal. A simple impact-effort matrix — scoring each opportunity by the hours it would recover against the effort required to implement — gives you a clear priority order. Start with the highest impact, lowest effort opportunities. These are your Quick Wins.

Weeks 5–6: Implement the Quick Wins. With two to three weeks remaining before 30 June, there is still time to implement one or two meaningful automation projects and have them running before the end of the financial year. Document collection and report assembly are typically the fastest to implement and the most immediately impactful for EOFY specifically.

30 June and beyond. The practices that use the EOFY window well don’t just survive it — they come out with a clearer picture of their operations and one or two fewer manual processes than they went in with. That compounds into the second half of the year.

The Honest Observation About Timing

Seven weeks feels like enough time to act on this. It is — but barely. The practices that will have automation working before 30 June are the ones that start the mapping exercise this week, not the ones that plan to look at it after the current wave of client work settles down.

In practice environments, things don’t settle down. EOFY ends and the next pressure cycle begins. The window to do something about this is open right now. It won’t stay open indefinitely.

If you’d like a clear picture of where your practice’s AI opportunities are — not in theory, but in your specific workflows — an AI Opportunity Assessment is designed for exactly this moment. A 90-minute session working through your operations, a written report within 48 hours covering your opportunities, the financial impact, and a prioritised AI adoption roadmap for acting on what we find. $1,500 excl. GST.

If you’d like to understand what’s involved before committing, I’m happy to start with a 20-minute discovery call first.

Reach me at team@quigly.com.au or visit quigly.com.au.

Mark Williams is the founder of Quigly AI, an AI consulting firm helping Australian SMBs discover and act on their AI opportunities. Based in Melbourne VIC, delivering nationally.