Paraic Burke, Head of Tax, PwC Ireland: In 2025, AI adoption is shifting from pilots to scaled production. Investment and performance have improved, and clearer regulation gives organisations confidence to embed AI. Surveys show productivity gains; cost savings lag where data and integration are weak. Ad hoc experiments rarely scale and durable impact comes from strong data, platform integration and robust governance.

The digitisation of tax compliance is becoming more standard as electronic mandates and AI redefine obligations. E-filing, APIs and Europe’s move to near real-time VAT and eInvoicing set the pace, while the first full year of Pillar Two raises the bar for data, controls and disclosure. Enterprise AI is now regulated, and production grade is compressing timelines and lifting expectations for accuracy, auditability and agility.

Leading groups are scaling where it matters. For Pillar Two, they’re transitioning to end-to-end GloBE calculations with transparent data lineage, automated classification of book–tax differences, and anomaly detection before period close. From an indirect tax viewpoint, AI embedded in determination validates invoices, infers appropriate VAT treatments from contracts and purchase orders, and flags misclassifications up front, boosting First Time Right processes and reduces remediation.        

The direction of travel is simple: standardise data, codify rules, automate the routine and focus humans on judgement. That shifts tax from constraint to an enabler, meeting new obligations at speed while informing deals, supply chains and operating models.

Compliance copilots are now more mainstream in tax return preparation and provision technology platforms. Using secure retrieval from libraries of guidance and law, they can draft workpapers, explain calculations and link to authorities for review, shortening cycle times while maintaining defensibility. In transfer pricing and controversy readiness, AI can accelerate screening of comparables, check consistency across intercompany agreements, invoices and customs data, and assemble master and local files. As part of the audit procedures, analytics can prioritise requests and anticipate focus areas from historical patterns, with oversight to avoid overreliance on probabilistic outputs. Across processes, foundation models can extract tax-relevant attributes from unstructured sources such as contracts, invoices, and customs forms.

What distinguishes 2025 for tax functions is maturity over novelty. Deployments are becoming narrower, better governed and tied to outcomes: faster close, fewer errors, stronger audit results and clearer control evidence. Successful implementations start with well mapped data, integrated systems and controls aligned to evolving regulations. As AI becomes built into tax operations through 2026, the differentiator will be professional judgement applied to better, faster information, supported by governance that keeps models current and compliant. The technology reached an appropriate level to deploy on scale; value now depends on disciplined execution and the human expertise that directs it.

This article appeared in the February 2026 edition of the Irish Tax Monitor.