Tax artificial intelligence technology is advancing rapidly. In just a year, KPMG plans to move from a chatbot to an advanced system that allows you to manage the entire process.
AI “Persona” – customer templates for various tax scenarios – help consultants work more efficiently. These tools drive advances in next-generation AI systems poised to significantly improve efficiency for consulting teams and tax departments.
The integration of human expertise and machine intelligence in tax work has begun.
Large AI
Today, AI models already handle most of the work of selected tax projects. They are useful by choosing legal rules, extracting data and doing initial analysis. For the rest of the work, data processing tools will be used to enable tax professionals to improve AI proposals by focusing on complex legal issues.
Large companies employ similar strategies, using AI to simplify large amounts of taxes and legal challenges. This approach requires only basic AI knowledge, but effectively streamlines complex processes.
One such example is in the German KPMG. It implemented an annual creation and automation for the update of local and master files for clients, a company in the European real estate sector.
The client has revealed that the generation AI is effective for 40% of tasks that have previously caused significant bottlenecks, allowing traditional labor-intensive transfer pricing documentation processes to be streamlined.
“Think Tank” Agent
The following year, KPMG observes the trend of AI in tax and legal services evolving beyond simple task automation and chatbots to more advanced “think tank” agents. These innovative agents work with an emphasis on achieving comprehensive goals.
Rather than just adhering to predefined instructions, these agents are equipped to autonomously manage the entire workflow. Those responsibilities include gathering information, interpreting the law, analyzing various scenarios, and drafting important documents.
Automating these processes allows experts to focus on complex issues that require human judgment. In addition to streamlining routine tasks, such agents leverage data analytics to identify emerging trends, assess evolving organizational processes against expected outcomes, and increase tax quality and legal decision-making.
However, achieving the necessary autonomy and accuracy in these domains does not simply rely on a wide range of Internet data.
Even a single percentage point error can have serious economic consequences, so agents must be built using carefully tweaked AI models with situation-specific tax and legal content. However, because integration and maintaining these datasets can be complicated, many organizations can seek support from expert partners to keep AI agents reliable and effective over time.
The demand for customized monitoring and knowledge has also created new roles. For example, KPMG has already assigned subject experts to function as “AI quality evaluators.” It is responsible for implementing partially automated but primarily manual human-driven checks to ensure the accuracy of AI responses.
Additionally, we expect “data and knowledge curators” to be collected, labeled and updated regularly by an AI job role.
With some initial workflows already in motion and tweaks now perceived as important for superior performance, there is likely to be widespread adoption of these professional agents in the second half of 2026. At the time, both regular operations and complex issues within the tax and legal department were managed with unprecedented accuracy and efficiency.
Turnkey Automation Targeting
Currently, AI automates some tax tasks, but heavy data tasks still require classic coded software. Because both AI prompts and classic software code are inflexible, new projects or workflow changes require major updates, suppressing the adoption of wider AI. Furthermore, this approach limits the ability of AI to learn and improve over time.
In contrast, agent-based models start with a broader range of purposes than fixed prompts, such as preparing specific tax documents. Use AI to learn from all available data, iterate through multiple tasks (including heavy data) and increase efficiency over time.
Early testing shows that this method significantly reduces the need for human monitoring and streamlines workflows.
The focus will be on AI agents development and support solutions until major technology providers provide a universal AI platform. Local AI models will improve, but cloud-based solutions may dominate. Organizations should focus on designing and scaling robust AI agents. It is important that agents have access to relevant knowledge and tools.
This article is based on Bloomberg Act, Bloomberg Tax, Bloomberg Government, or its owner, Bloomberg Industry Group, Inc. It does not necessarily reflect the opinions of the
Author information
Christian Tender is the director of tax and legal AI at KPMG International.
Eduard Seregin is Manager, AI and Product Management at KPMG Deutschland.
Written for us: Author's Guidelines
