AI in tax and audit: 32% time savings per task, 5x faster review, 80% return automation
Strong survey numbers, few named firms
The strongest public data on AI in tax and audit comes from industry surveys. The CPA.com / Blue J 2025 report, Thomson Reuters' user research and similar sources carry hard numbers. Named firms with audited, attributable outcomes are scarce, and that scarcity says something about where the industry is. With that on the table, here is the data.
Tax research adoption nearly doubled in a year
The CPA.com / Blue J survey found 60% of tax firm respondents use AI for tax research at least weekly in 2025, up from 33% the year before. When half the profession moves from occasional experimentation to weekly reliance inside twelve months, firms still treating AI as a pilot have a competitive problem.
Read it as a staffing and pricing story. If your peers are cutting research time by that much, your billing model and your associate workload assumptions are both under pressure.
32% per task
Thomson Reuters reported average time savings of 32% per task across tax and audit work among users of its AI tools. No firm names, but the sample is large enough to treat as an industry baseline.
32% compounds fast. A senior associate running 20 substantive tasks a week gets back roughly 6.5 hours. Across a busy season that is real capacity, either returned to judgment work or absorbed as volume growth without new headcount.
Return preparation and review
The CPA.com 2025 AI in Accounting report has the sharpest operational numbers in the public record. Some firms reported over 80% automation of individual return preparation in certain return types. The same report cites up to 70% reduction in time spent on manual tasks and 5x faster review cycles for tax prep and audit work.
These are self-reported outcomes from firms already running AI in production. The 5x review compression is the one for a managing partner: if review is your busy-season bottleneck and a peer firm moves through it five times faster, the capacity and margin gap is large.
Deloitte has disclosed GenAI and agentic capabilities in its audit platform for document review, with no measured outcomes published. Big Four AI investment is reported in the billions. The infrastructure bet is made.
What this means for firm leaders
The gains are real, and they are concentrated in firms that moved past experimentation into governed, production deployment. The 80% automation and the 5x review compression come from firms that connected AI to live workflows with enough process discipline to measure the result.
The risk is deploying without the audit trail. Tax and audit carry regulatory exposure, client confidentiality obligations and professional liability. An AI workflow that cannot show its reasoning, flag its confidence or produce a reviewable log of what it touched is a liability dressed as efficiency.
Before you build or buy: can you see what the system did, and can you defend it to a regulator or a client? If the answer is anything short of a clear yes, the efficiency isn't worth the exposure.
- Food and beverage supply chain AI: what PepsiCo, Nestlé and a pork producer measured$30M in additional gross margin identified by a single pork producer
- Wealth management AI: 85% call it the top compliance priority, 44% don't validate outputs44% of AI-using wealth firms have no formal validation of AI outputs
- AI in mining: four measured results from US Steel, BHP, Vale and FortescueVale: 30% longer haul-truck tire lifespan, $5M saved in one year
- https://tax.thomsonreuters.com/blog/how-do-different-accounting-firms-use-ai-tri/
- https://www.cpa.com/news/blue-j-and-cpacom-survey-finds-ai-adoption-among-tax-firms-has-nearly-doubled-one-year
- https://tax.thomsonreuters.com/blog/tax-industry-trends-2026-why-accuracy-is-the-new-standard-for-ai/
- https://www.wolterskluwer.com/en/expert-insights/ai-transformation-in-accounting-firms
- https://www.wolterskluwer.com/en/know/artificial-intelligence-tax-accounting
- https://vikasgoyal.github.io/industries/ata/accounting-tax-advisory-ai-2026-04-09.html
- https://www.thomsonreuters.com/en-us/posts/innovation/innovation-at-the-inflection-points-how-ai-is-transforming-tax-and-accounting/
- https://www.thomsonreuters.com/en-us/posts/innovation/the-real-ai-story-in-tax-and-audit-isnt-adoption-its-impact/
- https://www.certinia.com/blog/thomson-reuters-2026-ai-report-tax-audit/
- https://www.thomsonreuters.com/en/press-releases/2025/november/thomson-reuters-launches-agentic-ai-solutions-to-transform-tax-audit-and-accounting-workflows
- https://pubsonline.informs.org/doi/10.1287/mnsc.2022.04040
- https://www.thomsonreuters.com/en-us/posts/wp-content/uploads/sites/20/2024/07/Generative-AI-in-Tax-Firms-2024.pdf
- https://www.linkedin.com/pulse/big-four-spending-4-billion-ai-heres-which-accounting-bo-liu-vuruc
- https://www.thomsonreuters.com/en/institute/articles/future-of-professionals-report-analysis-tax-firm-talent-strategy
- https://tax.thomsonreuters.com/blog/the-impact-of-ai-on-the-tax-and-accounting-profession/
- https://www.wolterskluwer.com/en/expert-insights/the-power-of-ai
- https://www.linkedin.com/pulse/top-trends-2026-agentic-ai-reaches-tipping-point-tax-cpa-dvzse
- https://gitnux.org/ai-in-the-accounting-industry-statistics/
- https://www.cpa.com/sites/cpa/files/2025-06/2025_AI_in_Accounting_Report.pdf
- https://www.sciencedirect.com/science/article/pii/S1467089525000107
Fern's two-week Audit maps where a governed AI agent would pay off in your operation, and what has to be true to build it.
Book a scoping call →