Why Accounting Firms Aren’t Seeing AI Productivity Gains Yet

AI is everywhere in accounting right now. 

There is no shortage of new tools, bold claims, and headlines promising dramatic efficiency gains. But inside many firms, the reality feels much slower and less dramatic than the hype suggests. 

That doesn’t mean AI isn’t working. It means most firms are still doing the groundwork that actually enables it.  

AI tools

Most firms aren’t chasing every new AI tool. Instead, they’re focusing on how AI fits into their existing environment. 

At PVW Partners, COO Fiona McGill shared a similar approach, noting that while tools like Copilot are being rolled out, they’re doing so within a controlled environment. The team uses AI for data-driven tasks within their secure systems, with clear boundaries in place around usage.  

At the same time, firms are starting to formalise how AI is used day-to-day. Amanda noted that Power Tynan has built a large internal prompt library, now sitting at hundreds of examples, which are being embedded into templates and processes across the business. This helps move AI from experimentation into something repeatable and scalable.

AI vs Automation

One of the biggest misconceptions is that AI is the main driver of productivity gains right now. In reality, most firms are seeing far more impact from automation, especially when it comes to repetitive, rules-based work. 

This is playing out in very practical ways. Firms are using Robotic Process Automation to handle repetitive tasks like moving data between systems, pulling information from the ATO, and preparing workpapers. It’s not glamorous, but it’s delivering measurable time savings. Even then, automation isn’t removing people entirely from the process. 

This is where the real shift is happening: 

• Removing manual admin
• Reducing double handling
• Standardising processes
• Freeing up time for higher-value work 
  

Rather than replacing work, automation is changing the nature of it, reducing manual effort and creating space for higher-value thinking.

Team member buy-in

Even with the right tools and workflows, results don’t scale unless the team is brought along on the journey. 

At PVW Partners, Fiona McGill described how they’ve introduced monthly sessions where team members share what they’ve built or discovered. The goal is to avoid efficiencies being created in isolation and instead encourage a more coordinated, firm-wide approach.

It’s a simple idea, but it’s powerful. When people see what others are doing, it builds momentum. 

Amanda Kenafake shared a similar focus at Power Tynan but also highlighted the reality of managing change at scale.  

Across both firms, the pattern is clear: 

• Adoption doesn’t happen automatically.
• It needs structure, visibility, and ongoing reinforcement. 
  

Governance

Governance has emerged as one of the most important, and often underestimated, parts of AI adoption. 

Fiona McGill reinforced that governance doesn’t need to be overly complex to be effective. For firms starting out, she pointed to existing guidance from industry bodies but emphasised that policies ultimately need to reflect each firm’s own risk appetite and operating environment. 

Governance isn’t slowing firms down; it’s enabling confident adoption. 

When teams understand what they can and can’t do (and why), they’re far more likely to engage and use the tools properly.

Practical takeaways: what firms should focus on instead

So what does this mean in practice? The firms seeing real progress aren’t chasing hype. They’re focusing on fundamentals. That looks like: 

• Connecting systems so information flows properly
• Automating repetitive, rules-based processes first
• Embedding AI into existing workflows with clear policies, not bolting it on
• Bringing the team along through shared learning  
  

As one of the final insights from the session summed up: AI doesn’t fix broken processes - it amplifies them.

Or put another way: faster poor work is still poor work.

There’s also a broader mindset shift happening.

Automation increasing means that less time is spent on manual prep and instead on reviewing and interpreting the generated content. As a result, junior team members need to be trained differently.

And while it might feel like things are moving slowly, that’s often a sign the right work is being done.  

Final Word

It’s easy to feel behind when it comes to AI.

But one of the most reassuring insights from the discussion was this: most firms are in the same position - working through the fundamentals, testing what works, and building capability over time.

The firms seeing real results aren’t moving the fastest. 

They’re the ones focusing on foundations, connecting their systems, and bringing their teams with them every step of the way. 

And then layering AI in where it genuinely adds value. 

For further depth on these topics, watch the full webinar above.