For the last two years, finance and accounting teams have been told to adopt AI or get left behind. General-purpose tools are useful, but sales tax compliance requires systems designed for frequently changing, jurisdiction-specific rules.
Why general AI falls short
Sales tax is not simply a reasoning problem. The rules vary by state, product category, and transaction type, and they change often. A confident but incorrect answer can become an under-collected liability, a missed filing, or a state notice months after the transaction.
KPMG's Q2 2026 Global AI Pulse surveyed more than 2,100 senior business leaders. Only 7% reported established ROI from their AI investments, and organisations with strong visibility into AI operating costs were five times more likely to report it than those without (15% vs. 3%). These findings reinforce the need to match an AI tool to the task it is actually built to perform.
Purpose-built AI for sales tax
Kintsugi applies AI to high-volume, rules-based work such as nexus monitoring, product classification, address validation, and risk assessment. Tax calculation relies on rule-based logic, and people validate critical outputs and support customers on the decisions that carry risk.
That division of responsibility keeps AI focused on monitoring, flagging, classifying, and routing while precision-sensitive compliance work stays with systems and people built specifically for tax.
Questions to ask about an AI tax tool
- Do you understand the cost of each query as usage-based pricing expands?
- Is someone accountable for reviewing answers before they become filings?
- Does the tool track jurisdiction-specific rules, or is it relying on a general-purpose model?
Purpose-built does not mean less capable. It means the technology is matched to the narrow, high-stakes work of sales tax compliance.
