Highlights
- 65% of tax professionals now use GenAI tools, yet only 26% measure the resulting ROI.
- Three years of data show adoption came first, strategy came next, and measurement remains the biggest obstacle.
- Indirect tax faces higher stakes from shallow AI adoption due to its high transaction volume and tight deadlines.
Ask an indirect tax professional whether they personally use AI, and most already say yes. 65% now use publicly available GenAI tools like ChatGPT for work. That statistic alone used to be the headline. Now it’s barely the starting point.
For years, Thomson Reuters has tracked how AI is reshaping how professionals work. Each year, the question professionals were being asked got harder.
- In 2024, it was simple: Should we adopt this?
- By 2025, it had sharpened: Do we have a strategy?
- This year, in 2026, the question is the one that separates functions that are ready from those that only look ready: Are you measuring what you’re doing, and are you going deep enough?
Indirect tax now has its own answer, and it’s a more complicated one than the adoption number suggests.
Jump to ↓
2024: Adoption gained momentum
2025: Strategy became the dividing line
2026: The data caught up, and it’s a mixed picture
Where the data about AI in corporate indirect tax leaves you
2026 AI in Professional Services Report
GenAI is here, agentic AI is coming — and business model shifts are next
Access report ↗
2024: Adoption gained momentum
When generative AI first reached professional services, indirect tax moved fast, and for good reason. Rate changes across jurisdictions, e-invoicing mandates, and real-time transaction reporting had already stretched teams thin before AI ever entered the picture. Professionals across tax told us they expected AI to free up nearly 200 hours a year and believed it would have a transformational impact on their work. Skilled labor shortages were straining tax departments more than legal or risk, fraud, and compliance. For a function drowning in rate tables and invoice exceptions, AI looked like relief.
That instinct hasn’t changed. What’s changed is what “adoption” means.
2025: Strategy became the dividing line
By 2025, it was clear that using AI and having a plan for AI were two different things, and only one of them showed up in the numbers. Organizations with a visible AI strategy saw roughly double the revenue growth and three-and-a-half times the benefit of those adopting tools informally. In-house departments with their own strategy, rather than just a borrowed one from the broader organization, captured more of that value than the ones waiting for direction from others.
For indirect tax specifically, this was the year the divide was felt more than it was measured. Some teams were clearly using AI to triage invoice exceptions or pre-screen nexus questions before they hit a human reviewer. Others were still treating AI as a research shortcut and nothing more. The data to prove which approach was providing an advantage wasn’t there yet.
2026: The data caught up, and it’s a mixed picture
This year, for the first time, we have precise numbers on where tax really stands, and it’s a genuine test of depth, not just adoption.
Adoption is strong. 65% of tax professionals personally use publicly available GenAI tools, and tax research now ranks as the single most common use case in the profession, followed closely by document review and return preparation. Translate that into indirect tax terms and it means AI is already sitting inside VAT and GST research, invoice review, and return filing at a majority of organizations.
Despite that, in-depth adoption hasn’t kept pace with general adoption. Only 18% of tax organizations have moved into agentic AI, the kind that can independently research a jurisdiction’s rate change, flag the affected transactions, and draft the correction, rather than simply answering a single prompt. Only 26% of tax departments say they measure the return on that AI investment at all.
That last number matters most. Organizations with a formal AI strategy now drive three times the ROI of an ad-hoc approach. However, you can’t capture or prove a return you’re not measuring. Across the entire profession, 82% of organizations either aren’t collecting ROI metrics on AI or don’t know if they are. Indirect tax has gone the path of least resistance (general adoption) while the path that delivers the most value (in-depth adoption and measurement) has yet to be embarked upon for most.
Why this AI discrepancy hits indirect tax harder
The discrepancy between wide adoption and shallow depth matters more in indirect tax than almost anywhere else in tax. Indirect tax runs on volume. It includes thousands of transactions, dozens of jurisdictions, and filing deadlines that are measured in days, not months. A function operating at that scale without agentic tools or ROI measurement leaves efficiency on the table and isn’t measuring value in the one area of tax where errors cascade and pile up the most.
Where the data about AI in corporate indirect tax leaves you
None of these insights mean indirect tax is behind. Strong personal adoption, especially in research and return preparation, is a real advantage. It does mean the next competitive advantage doesn’t lie in asking yourself do we use AI? but rather do we know what it’s doing for us, across every jurisdiction we file in, and are we willing to let it do more?
See exactly where corporate tax stands against the rest of the profession by downloading the 2026 AI in Professional Services Report.
