From adhoc AI to accountable tax workflows 

AI is now part of the day job in tax. That’s the new reality many teams are grappling with at pace: AI is being used across research, drafting, document analysis and internal knowledge tasks — often at the same time as firms are trying to tighten standards, reduce risk and improve client transparency. 

That tension sits at the heart of Tolley’s AI workflows in tax report, based on a February 2026 survey of 400+ UK tax professionals. The findings show a profession moving quickly on day-to-day usage, but still catching up on the operating model needed to make AI reliable at scale. 

High usage doesn’t automatically mean high maturity 

AI has become normalised in core areas of tax work. For example, 84% of practitioners use AI for tax research, with strong usage also reported in knowledge management (68%) and document analysis (67%). 

Yet only 23% say AI is embedded in their team’s strategy and operations. 

That gap is where risk creeps in. If individuals are left to choose tools and develop “how we do AI” informally, outcomes become inconsistent. Review standards can vary by manager. Accountability becomes blurred. And it becomes harder to explain to clients what controls sit behind AI-assisted work. 

An adhoc AI workflow strategy won’t raise standards. It can quietly lower them. 

Trust is the real differentiator in 2026 

The report signals a clear preference for tax-specific AI tools for core tax work, and the reasons are practical: confidence in sources, consistency of outputs, and the ability to evidence what the tool is doing. 

As Paul Aplin, Deputy President of the Chartered Institute of Taxation, puts it: “Trust in AI is built on assurance of reliability, accuracy, consistency and explainability.” 

That emphasis on trust also shows up in the risk concerns practitioners continue to report — particularly around inaccurate or fabricated outputs, over-reliance, and data confidentiality. In other words, “AI use” is not the same as “AI you’d confidently stand behind”. 

Redesign the workflow, not just the toolkit 

The most useful shift leaders can make now is moving from tool-led experimentation to workflow-led design. That means being explicit about: 

• Approved use cases: what AI should and shouldn’t be used for in your team.  

• Approved tools: when a tax-specific tool is required for core work.  

• Minimum review standards: what must be checked, by whom, and how it’s evidenced.  

• Clear accountability: who owns the final output when AI contributes to deliverables.  

• Security controls: ensuring sensitive data is handled in an appropriate, ringfenced environment.  

Done well, AI can accelerate the first phase of tax work — while strengthening quality checks and supervision, rather than weakening them. 

What to do next 

AI workflows in tax makes one point hard to ignore: the next phase of AI maturity won’t be defined by who has access to tools. It will be defined by who can embed AI into delivery with governance, consistency and transparency. 

To explore the full findings and what they mean for tax teams, read Tolley’s AI workflows in tax report. 

Meet the Tolley team on stand 580 at Accountex London, taking place at Excel on the 13-14 May 2026.

Register for your free ticket here.

The post From adhoc AI to accountable tax workflows  appeared first on Accounting Insight News.