How AI Upskilling Translates to Tax

As more companies and business groups pursue AI enablement, human resources (HR) leaders are devoting more time and thought to developing the new skills that organisations need to maximise the returns on their investments in intelligent automation. These strategies are relevant to tax executives, who are just as eager as other business leaders for fresh insights into AI upskilling.

“An intentional, outcomes-led approach to continuously replanning, reskilling and redeploying talent – grounded in dynamic skills intelligence and supported by a culture of AI enablement – drives sustainable performance gains,” according to Mercer’s Global Talent Trends 2026 report, which identifies AI-driven redesigns of workflows and job roles as one of four strategic objectives that HR groups are pursuing this year.

The report’s authors discourage HR executives and other business leaders from focusing on how AI “could be applied to existing jobs and processes to eliminate human work.” Instead, business leaders should focus on what work needs to be performed. With that objective in mind, they can then deconstruct job roles and workflows to determine which tasks can be “substituted, augmented or transformed using AI.” Mercer also emphasises that AI generates value “when work is redesigned around it, not when technology is layered onto outdated work models and processes.”

That makes sense. I’d add that AI’s value derives not by how autonomous it becomes but by how effectively it amplifies and scales human expertise. As I’ve pointed out, tax groups that use AI to handle complexity and surface what truly matters give their tax professionals more bandwidth to focus on judgement, strategy and higher-order problem solving.

In addition to assessing traditional workflows and job roles, tax leaders should also consider how to equip their teams with the AI skills they need to thrive in an increasingly automated era. Four AI-upskilling areas to consider include:    

• Prompt engineering: This basic level of training is also foundational. It instructs tax professionals on how to interact with generative AI tools to generate the most useful responses. While generative AI interfaces resemble traditional search engine windows, developing the ability to create structured, context-rich prompts is much more involved than performing keyword searches by inputting five terms. A gen-AI prompt that contains 1,000 words of well-designed context will generate far more illuminating and actionable responses than a few questions.
• Low-Code, No-Code and Vibe-Coding Tools: Microsoft Copilot Studio, Oracle AI Studio and similar platforms allow users without extensive programming knowledge to automate processes by visually connecting components. No-code programming relies on visual tools to program while low-code programming requires some basic scripting expertise. Vibe coding, which is getting more attention these days, leverages natural language prompts so that non-coders can program.
• Semantic Knowledge of Data: A deep understanding of business data sources and the semantic meaning of data elements enables users to apply AI with richer context and clearer intent. This knowledge helps teams instruct AI systems more effectively to achieve desired business outcomes. Reskilling in this area can also include curating and maintaining data quality, integrity and cleanliness to ensure that AI-driven automation is reliable and trustworthy.
• AI Capability Tracking: This skill area focuses on staying informed about advances in AI and understanding how they apply to the business and the tax function specifically. It involves separating practical capabilities from market hype as well as identifying what is possible today versus what may emerge in the future. Ongoing capability tracking can also serve as an input into identifying additional skills and training needs over time.

As tax leaders enhance their focus on AI-driven workflows and related roles redesigns, they should recognise the value of amplifying the skills of the humans overseeing the agents and automation that perform more tasks.