Six AI reflections and 2026 outlook for professional services firms
As 2025 comes to a close, BDO's national digital leader, Nick Kervin, and digital partner, Fahim Khondaker, reflect on AI use in the professional services sector.
The professional services sector continued to experiment with artificial intelligence (AI) in 2025. Many of the leading organisations are now facing the need to confront the structural changes required to turn those experiments into meaningful productivity gains. As we close out the year, we highlight six key reflections on AI for the sector:
1. There are lots of AI events and seminars
A key highlight for this year was when an event organiser called three weeks prior to a function, requesting to cancel our AI keynote presentation because the audience expressed that they had enough discussion on AI and would rather a presentation on entrepreneurship or leadership.
This was a significant lead indicator, suggesting that the awareness phase in the new tech’s lifecycle is now mostly over, and a need to add significant practical value to audiences now exists.
2. Leadership alignment is key
Almost all organisations are misaligned from the CEO down with respect to AI. There is a significant knowledge gap in understanding existing AI and the recent emergence of generative AI (GenAI). Moreso, there is no alignment on the organisation’s appetite and approach to adopting AI, resulting in ad hoc adoption across pockets of the organisation.
Unlike other projects, AI adoption needs more than just one or two executive sponsors – it needs alignment across the entire leadership team on purpose, relative importance, funding and implementation approach.
3. We are in the adoption phase, and there are policy, regulator, and economic factors at play
There is a significant focus on adoption now. The National AI Centre has released new adoption guidelines, and Australia released a new National AI Plan under the headings of “Opportunities, Benefits and Risks”.
The benefits all focus on the adoption of AI at scale across every business in every sector. The opportunities component relates to capital investment in data centres and other infrastructure, while the risks part outlines the key societal risks we must be aware of as we venture into this new era as a nation.
Policymakers, regulators, industry and the public will all have to grapple with the various trade-offs and implications of the various factors in play.
4. AI capability is improving, but not quite there yet
As expected, GenAI models have improved significantly to the point where the reasoning models can handle far more complex instructions and produce detailed outputs in the form of presentation decks, reports, videos, posters, podcasts, and detailed financial models.
Everything still needs a lot of validation, and the error rates are still higher than tolerable to allow us to delegate everything to AI; however, the models are infinitely better now than they were when ChatGPT was released in 2022.
5. Operating models, value chains and processes
Most organisations are trying to implement AI into their existing business models and processes. While this is a good starting point, many prominent thought leaders on this topic are of the view that we have to reimagine our business models around AI instead.
Using AI to find efficiency in our existing businesses (which appears to be the primary reason most organisations are looking at AI) will have a natural ceiling of 25–40 per cent.
6. Leveraging your own intellectual property and institutional knowledge
Everyone is either using or will soon be using AI. The competitive advantage and differentiator will come from how well we use it and how well we train it using our own organisation’s intellectual property and institutional knowledge.
After a decade in which formal knowledge management fell off the strategic agenda, its relevance is being rediscovered almost overnight. The race to capture, organise and govern internal knowledge will accelerate sharply in 2026. Businesses will invest heavily in revitalising knowledge systems that had previously been left to age quietly in the background, recognising that the future competitiveness of their AI tools depends on it.