AI in tax: from early promise to everyday reality

Ahead of ITR’s AI in Tax Forum in London on Wednesday, April 29, we spoke with WTS Germany’s Tim Zech and keynote speaker Jeff Soar of WTS UK about AI’s potential to transform the industry.

To hear more from WTS experts on the future of AI and innovation in tax, register for the event here.

Tim: We’ll stop talking about “AI projects” and start talking about “how we work.” In two to three years, AI will be embedded in many routine tax processes: drafting responses to tax authority queries, preparing standard memos, consolidating data for compliance, or running first‑pass risk checks. The big shift is that the default will be: “Can the system do the first 80%?”, while the human focuses on judgement, escalation, and sign‑off.

Jeff: I agree. The most visible change will be the speed of service. Tasks that today take several days – like multi‑jurisdictional research for a business model change – will be compressed to hours, because AI will pre‑filter sources, summarise positions, and highlight conflicts. And tax authorities will expect that level of speed on the taxpayer side, because they’ll increasingly use AI themselves for risk scoring, anomaly detection, and case selection.

Jeff: Three stand out:

Copilots that read legislation, case law, guidance and internal precedents, then draft a first version of a memo, ruling request, or audit response. The tax professional curates, corrects, and signs off, but the “blank page” disappears.

Tools that can read contracts, invoices, and enterprise resource planning (ERP) extracts to flag potential indirect tax issues, permanent establishment indicators, or transfer pricing red flags. This will be especially powerful in controversy and due diligence.

AI that can simulate the tax impact of business decisions – supply chain changes, IP migrations, new booking models – drawing on structured tax rules, rates, and your own historical data.

Tim: I’d add a fourth: knowledge management. Many tax functions are sitting on decades of memos, rulings, notes, and email threads that only exist in people’s heads or in shared drives. Turning that into an AI‑navigable knowledge base so colleagues can ask, “Have we seen this before?” and get a curated answer will be a game changer for quality, consistency, and onboarding.

Tim: We’re heading toward much more data‑driven and transparent relationships. Authorities will increasingly use AI to cross‑check declarations against other data sources, like customs, payroll, financial statements, and even public information. That means fewer random audits and more targeted questions. For companies, being “AI‑ready” means having clean, well‑structured tax data, a clear documentation trail, and the ability to explain how positions were reached, especially where AI tools were involved.

Jeff: This will also change the adviser–client dynamic. Advisers will be expected to bring AI‑enabled services, not just human expertise. That includes:

• Helping clients design and govern AI use in their tax function;

• Offering AI‑accelerated reviews, controversy support, and diagnostics; and

• Co‑creating tools and knowledge assets that clients can use day‑to‑day.

The competitive edge for advisers will be the combination of deep technical tax knowledge and the ability to embed that knowledge into robust AI workflows.

Jeff: One misconception is that “AI will replace tax professionals.” In reality, AI will replace certain tasks, particularly repetitive and pattern‑based work. But tax is full of ambiguity, judgement, negotiation, and ethics, areas where human responsibility remains central. The risk is not replacement; it’s complacency – assuming the machine is right because it sounds confident.

Tim: Another risk is underestimating governance. Many organisations are still experimenting with generic AI tools on highly sensitive data. In tax, that’s a serious issue. Over the next two to three years, we’ll see:

• More scrutiny from regulators on how AI is used in decision‑making;

• Stricter internal policies on data usage, model selection, and documentation; and

• A need to demonstrate that AI‑assisted positions are explainable and defensible.

The winners will be those who move fast and build strong guardrails from day one.

Tim: A good AI strategy for tax is practical, prioritised, and governed. In concrete terms:

For example: response times to audits, quality and consistency of advice across countries, bottlenecks in compliance, or reliance on a few key experts.

Don’t try to “AI‑enable everything.” Show value quickly, then scale.

Without structured tax data and curated knowledge bases, AI will be shallow and unreliable.

Clear rules on what AI may and may not do, how outputs are reviewed, and how decisions are documented.

Jeff: I’d add: plan for evolution, not just implementation. Models will improve, regulations will tighten, and internal expectations will rise. Build a roadmap that anticipates:

• Integration with other enterprise systems (ERP, e‑invoicing, document management);

• Growing collaboration with IT, legal, and risk; and

• Gradual redesign of processes as AI becomes more capable.

An AI strategy isn’t a one‑off project; it’s a capability you’ll be continuously refining.

Jeff: We’re seeing three layers emerge:

1. Core tax professionals who are comfortable working with AI – prompting effectively, reviewing outputs critically, and knowing when to challenge or discard what the system suggests.

2. ‘Tax technologists’ who understand both tax and data/technology – people who can translate tax logic into rules, workflows, and model requirements, and who can talk to IT and data teams.

3. Specialists in governance and risk who ensure that the use of AI aligns with regulatory expectations, internal policies, and ethical standards.

Not everyone needs to be a coder, but everyone will need to be ‘AI‑literate.’

Tim: From a leadership perspective, the key skill is orchestration. Heads of tax will need to:

• Set the vision for how AI changes their operating model;

• Prioritise investments and manage change in their teams; and

• Work closely with the C‑suite on the tax implications of broader AI initiatives.

Those who embrace this role will position tax as a strategic partner in the organisation’s digital journey.

Tim: I’d recommend three very pragmatic steps:

Even a simple “what’s allowed, what’s not, and how we review outputs” is better than silence. It reduces risk and opens the door to structured experimentation.

Choose use cases where the risk is manageable and the benefit is visible, e.g., AI‑assisted research or document summarisation. Use them to learn what works, what governance you need, and how your team responds.

Curate key memos, rulings, and guidelines; define ownership; and think about how that content could be exposed to AI securely. This improves your function today and prepares you for more advanced tools tomorrow.

Jeff: I’d emphasise the learning agenda. Encourage your team to experiment within defined boundaries. Build a small interdisciplinary group – tax, IT, data/privacy, risk – to oversee AI in tax. The goal isn’t perfection; it’s building capability and confidence so you can move faster when the time is right.

Jeff: Forums like this are critical because no one has all the answers yet. We’re collectively figuring out best practices, guardrails, and what “good” looks like.

Tim: WTS sees its role in three ways:

• Thought leadership: helping define what responsible, effective AI in tax looks like;

• Co‑creation: working with clients to design and implement AI‑enabled processes and solutions that fit their reality; and

• Community building: using platforms like this forum to connect tax leaders, share experiences, and accelerate the learning curve for everyone.

Over the next two to three years, the organisations that succeed with AI in tax will be those that combine ambition with responsibility. Our goal is to help make that possible.

The next few years will not be about whether AI matters in tax – it clearly does. The real differentiator will be how tax functions and advisers integrate AI into their work: thoughtfully, transparently, and with a clear focus on value and accountability.

The ITR AI in Tax Forum will be a space to explore exactly that, and to turn vision into concrete next steps.