In indirect tax, you rarely get to ask whether your technology is the right shape for the job, because you’re busy using whatever gets this month’s returns out the door, managing endless audits and supporting business priorities that were due yesterday.
I’m Nikki Flom, I recently joined Fonoa from Meta, and I spent my days on that side of the desk. Attending TEI Annual National Conference and ITR’s AI In Tax Forum in the US gave me a rare opportunity to step away from the day-to-day and hear how other tax professionals are approaching AI.
For a few days nobody needed me to deliver anything, so I could sit back and listen to the whole profession argue about AI. Six takeaways stuck with me.
1. Almost everyone is still figuring AI out
In-room polling at ITR showed that around 75% of tax teams are still in the early stages of exploring generative AI, and fewer than 40% feel ready for AI agents.
If you’ve felt behind, the room suggests you’re in good company. The risk considerations are the spotlight. The AICPA’s AI risk framework maps out 20 risk areas, and its new Council on AI Risk in Tax is there to help teams weigh them.
The best advice I heard was to start small, keep your team in the loop, and test outcomes together.
2. Indirect tax has the right shape for AI
The rules repeat, and the same transaction has to land in the same return box every month. That predictability makes indirect tax a good first place for AI to show what it can do.
Cleaner data and fewer manual steps give the team time back for the strategic work that rarely gets room on the calendar.
3. Let AI do the reading and let rules do the math
Even in indirect tax, you need to know where to draw the line.
Generative AI is great for research and drafting. Validation, calculations, and tax rates need deterministic logic, where the same input always produces the same answer.
We’ve all seen AI deliver a wrong answer with total confidence, which is why reliable, cited sources and the right time frames matter so much. A tax rate with no effective date attached is a guess.
4. Clean data is the unglamorous foundation
This one resonated with me more than any other. Clean data rarely gets the exciting slide, yet it and automation sit underneath every other idea on this list.
An AI tool can only be as reliable as the data invoices it reads.
5. Tax authorities are already using AI
While many of us debate whether we’re ready, tax authorities are investing in AI for audit selection, analysis, and fraud detection.
Expanding e-invoicing and reporting mandates around the world are also pulling indirect tax audits toward real-time direct exchange of transaction information, and shorter audit cycles.
The time between a transaction and a question about it keeps shrinking, and waiting means meeting those questions with yesterday’s tools.
6. Somebody’s name still goes on the return
Some of the liveliest conversations were about how AI and people share the work.
Who owns AI in the tax function? Who stands behind something an AI wrote?
The tax authorities were clear on this one: professionals must review the outputs and hold the final responsibility.
Where this leaves me at Fonoa
When I look back at what I wanted from tax technology in the day-to-day, I landed on four things: automation for the repeatable work, fixed logic for every number, a cited source for every analysis and records I could hand to an auditor with full confidence.
That’s the thinking behind Fonoa’s TaxOS for autonomous tax, which brings validation, tax determination, e-invoicing/re-reporting, and returns together under one reliable audit trail while keeping human judgment at the center. It’s a big part of why I joined.
If you’re interested in learning more, contact our team.
