The 98% figure — and why it needs unpacking

When Xero commissioned research through Cebr into AI adoption across UK accounting practices, the headline finding was remarkable: 98% of UK accounting practices already use AI in some form. (6 Figure Bookkeeper)

This sounds like near-complete transformation. The reality is more textured. “Using AI in some form” covers a spectrum that runs from a firm using an AI-powered spam filter on its email server, to one that has redesigned its entire client engagement process around AI-native workflows. The majority of practices sit much closer to the former than the latter.

What is unambiguously true is that AI is embedded in the accounting software stack that almost every practice uses. The major platforms — Xero, Sage, QuickBooks, and FreeAgent — have all embedded AI features into their core products, meaning that any practice using current versions of these tools is, technically, using AI whether they know it or not.

Making Tax Digital for Income Tax: the April 2026 deadline

The accounting profession's single biggest compliance event of 2026 is the launch of Making Tax Digital for Income Tax Self Assessment (MTD for ITSA) in April 2026.

MTD for ITSA requires self-employed individuals and landlords with income above £50,000 (from April 2026, reducing to £30,000 from April 2027, and £20,000 from April 2028) to:

For accounting practices, this represents a significant workflow change. Clients who previously provided annual records in January will now need to maintain digital records and submit quarterly data. Practices that have not already migrated clients to cloud accounting software will face a difficult period of accelerated transition.

AI tools are central to making this transition manageable. Xero's “Just Ask Xero” AI assistant aims to auto-reconcile more than 80% of bank transactions automatically, (6 Figure Bookkeeper) dramatically reducing the manual bookkeeping load that quarterly reporting creates. The practice that can bring its entire MTD-affected client base into an AI-assisted workflow is significantly better positioned than the one still processing manual records.

Platform AI: what Xero, Sage, and QuickBooks are actually doing

The accounting software platforms have been investing heavily in AI capabilities, and the features are now live rather than promised:

Xero: “Just Ask Xero” (JAX) uses conversational AI to answer bookkeeping questions, auto-categorise transactions, and guide users through reconciliation. The 80%+ auto-reconciliation target is its most significant capability claim. Xero also uses AI for cashflow forecasting, late payment prediction, and automated financial reporting.

Sage: Sage Copilot, embedded in Sage 50 and Sage Intacct, provides AI-assisted invoice coding, expense matching, and predictive financial analysis. Sage's “State of the Nation” report, however, reveals a stark disconnect between platform capability and user confidence: only 27% of UK accountants describe themselves as “very confident” in their firm's AI awareness, and just 23% receive any AI training from their employer. (AccountingWEB)

QuickBooks: QuickBooks AI provides transaction auto-categorisation, anomaly detection, and natural language financial queries. Its integration with the broader Intuit ecosystem includes TurboTax AI for tax preparation support.

The disconnect between platform investment and user capability is the defining tension in accountancy AI adoption. The tools exist; the knowledge to deploy them effectively does not.

The Jeremy Hunt controversy — and what ICAEW said

In 2025, then-Chancellor Jeremy Hunt made a remark that generated significant professional reaction: he suggested that young people might consider avoiding a career in accountancy because of AI automation risk.

ICAEW pushed back firmly. The Institute's position — which it published explicitly — is that accountants are embracing AI and securing their future, not facing displacement from it. (ICAEW)

ICAEW's argument is not that AI doesn't change accounting — it does. It is that AI changes what accountants do, not whether accountants are needed. The tasks at risk — data entry, transaction matching, basic VAT returns — have never represented the highest-value work in the profession. What remains is the judgment, the client relationship, the advisory capability, and the regulatory expertise that AI cannot replicate.

The Xero/Cebr research supports this framing: AI adoption has contributed to a £338 million profitability uplift across UK accounting practices — suggesting that AI is expanding the economic value of the profession, not contracting it. (6 Figure Bookkeeper)

The training gap: the profession's urgent problem

Despite the 98% AI penetration figure, the training data reveals a profession that is using AI tools without adequate understanding of what they are doing.

Sage's survey data is striking:

(AccountingWEB)

This creates significant risk. An accountant who is auto-reconciling 80% of transactions using Xero AI without understanding how the system categorises entries, what its error rate is, and how to identify miscategorisations is not using AI safely — they are delegating to AI blindly. In the context of HMRC submissions, tax returns, and financial statements, that blind delegation can result in material errors with significant consequences for clients.

The profession's CPD infrastructure has not kept pace with AI deployment. ICAEW, ACCA, and CIMA all offer AI-related resources, but the volume and accessibility of practical AI training for practising accountants — particularly in small and medium practices — falls well short of what the pace of platform development demands.

What AI cannot yet do in accountancy

AI transformation of accountancy is real and accelerating, but several critical functions remain human-dependent:

Judgment on complex transactions: AI can categorise straightforward transactions reliably. It struggles with complex, ambiguous, or novel transactions that require interpretive judgment — a factor that is particularly important in M&A accounting, group restructuring, and unusual capital transactions.

Client advisory conversations: The highest-value conversations in accounting — tax planning, business sale preparation, cash flow strategy, succession planning — require the relational trust and contextual understanding that AI does not replicate. Clients experiencing business challenges want a trusted adviser, not an algorithm.

Regulatory interpretation: Tax law changes frequently, and HMRC guidance evolves. Applying new rules to specific client circumstances requires current knowledge and interpretive skill. AI training data has a cutoff date and cannot reliably navigate legislative developments without human oversight.

Professional responsibility: The accountant's signature on a tax return or financial statement represents professional accountability. AI tools cannot bear professional responsibility. The human adviser who reviews and approves AI-generated work remains fully accountable for its accuracy.

Practical steps for accountancy practices in 2026

Immediate priorities:

Medium-term priorities:

  1. Assess AI tools for tax research, document drafting, and client reporting

  2. Consider AI-assisted advisory tools that can help surface insights from client financial data for business advisory conversations

  3. Update your engagement terms to address AI use and data processing

Key statistics at a glance

98% of UK accounting practices use AI in some form (6 Figure Bookkeeper)

Making Tax Digital for ITSA launches April 2026 — requiring digital records and quarterly submissions from self-employed above £50,000 income

MarGen specialises in content strategies for B2B professional services in the AI age. Get in touch.