Lesson 2: From isolated pilots to a future-proof tax function by 2030
The TFO Survey shows that tax and finance functions worldwide struggle to maintain a sustainable AI strategy. Not because of a lack of will, but because reality is tough. Every day brings new AI news: a new model that promises to change everything, a new feature you must test, or a new tax-AI startup flooding your inbox with demos and promises. The result? You’re unconsciously pushed into short-term thinking. You run from experiment to experiment, chasing the latest shiny object, making it hard to stay on course.
But when you ask tax directors to look ahead—not to next week, but to 2030—the conversation changes completely. It’s no longer about today’s tool but about a tax function deeply integrated into the business, making faster, smarter decisions because analyses are almost real-time, and where many repetitive processes are highly automated. You hear visions of small, agile teams supported by robust AI systems that process data, detect signals, and perform preparatory analyses. The future tax directors describe is about growth, new ways of working, and a tax function that consistently delivers more value to the organization.
Four Futures sessions
To help tax functions escape short-term mode, we use a practical approach that is both strategic and tangible. A key component is our Four Futures sessions. These aren’t theoretical models but a way to pull tax teams out of the here-and-now. Together, we make a mental leap to 2030: what does your tax function look like then, what role do you play in the organization, which activities are automated, and what skills do you need? From that future vision, we translate what you can do today to be robustly prepared—regardless of which AI-scenario becomes reality.
In parallel, we organize EY.ai Tax Lab sessions, where we take tax teams out of their daily routine for a full day and put them in experiment mode. There, we don’t just build strategic plans but real AI agents. In almost every session, similar themes emerge. Many teams start by building a knowledge or policy assistant to handle frequently asked business questions. Often, a horizon-scanning agent follows, monitoring tax authority websites, news sources, and case law, automatically translating them into organizational impact. And nearly every Dutch tax function wants to tackle document analysis: automatically reading, comparing, and matching the right fields from tens of thousands of contracts and invoices—especially for VAT purposes. These use cases help tax teams move from abstract AI ideas to concrete value.
Wild growth
Some organizations come to us with the opposite problem: they already have ten, fifteen, or even twenty GenAI initiatives running across different countries or departments. Then a new risk emerges: chaos. Pilots aren’t completed, quality varies by team, there’s no central governance, and no one can clearly explain the total business case. To break that cycle, we use the AI Factory model. This model helps tax functions move from scattered initiatives to a structured way of building. You work with reusable prompts to build agents, set up system integrations, and keep knowledge sources current. Governance, quality, and Responsible AI are central in the AI Factory model, and you assess value and ROI not per agent but across the entire portfolio. This creates calm and structure—and lets you scale much faster without losing control.
The key takeaway: don’t overestimate AI in the short term. It’s natural to be tempted by the latest model or feature, and yes, AI still makes mistakes—that’s part of learning. But don’t underestimate AI in the medium to long term. The changes tax functions will undergo in the next five years are huge, and to be ready, you need to make strategic choices today. Not to eliminate all uncertainty, but to maintain direction while technology races ahead.
Lesson 3: Without people, no sustainable AI transformation
If data is barrier number one, people are barrier number two. The TFO Survey shows that a lack of AI skills is one of the biggest obstacles to further adoption of AI in tax and finance. In practice, employees often go through an emotional curve with GenAI. First comes the magic: the first time someone uses a GenAI tool and gets a lengthy memo or complex email back in seconds, the reaction is often enthusiastic. It feels like you’ve gained a digital junior—always available, never tired.
Then comes the inevitable disappointment. People discover that AI sometimes gets things wrong, invents facts, misses context, or struggles with local nuances. Without clear guidelines, sentiment can flip: fun toy, but I can’t trust it, or I’m spending more time checking than doing it myself. Without guidance, the initial energy drains away, and GenAI becomes an occasional tool for a small group of enthusiasts.
With the right guidance, you reach a third phase: purposeful use. Colleagues understand what GenAI is good at and what it isn’t, how to craft effective prompts, what checks and balances are needed, and how to embed AI logically into existing processes and controls. They stop seeing GenAI as a magic solution for everything and start using it as a powerful tool for specific tasks.
Mindset, skillset, toolset
The faster you help your tax team through this cycle, the greater the chance AI truly takes root in how you work. That means investing as much in mindset and skillset as in toolset. The TFO Survey makes this clear: tax and finance leaders now consider data and technology skills more important for the tax professional of the future than pure technical depth. That’s not a license to neglect tax expertise, but it does mean the winning profile in 2030 combines tax knowledge with digital and analytical skills.
Leading organizations offer training that goes beyond “how the tool works” and covers bias, hallucinations, source verification, privacy, and Responsible AI. Teams include more diverse profiles: tax specialists with data affinity, data analysts interested in tax, and people bridging tax, IT, and business. In many cases, AI champions or ambassadors are appointed—colleagues given extra time and mandate to experiment, collect best practices, and bring others along.
