I read a lot of AI reports. Most of them say the same things: adoption is accelerating, firms need a strategy, the future is here. Fine. But what I actually want to know is what firms are doing. What did they build? What did it replace? What happened?
The CPA Trendlines 2026 Cornerstone Report on agentic AI is different. It’s dense, it’s long, and it’s full of real examples from named firms. I pulled out the case studies because that’s where the interesting stuff is.
First, the headline numbers
AI adoption in accounting firms jumped from 9% to 41% in a single year. That’s not a gradual trend — that’s a threshold being crossed. Firms with a clear AI strategy are 3–4 times more likely to see measurable gains than those without one. And 35% of professionals say they’re using AI tools daily.
The report calls it a tipping point. I think that’s right. The question now isn’t whether to adopt — it’s whether you understand what agentic AI actually means in practice.
Quick definition: agentic AI doesn’t just analyze or suggest — it acts. It autonomously performs tasks within parameters you set. It’s the difference between a tool that surfaces information and one that logs into 100 accounts, reconciles the data, flags anomalies, and hands you a report.
What firms are actually doing
Tax: RSM’s internal AI for complex memos
RSM US built an internal generative AI called Guidepath specifically for writing complex tax position memos. When a client has a thorny issue — interpreting a new law change, building a position for an unusual transaction — Guidepath searches the firm’s internal knowledge base of prior memos, court rulings, and IRS releases, then drafts a detailed position paper. Senior tax experts review and refine rather than write from scratch.
It also works as a “resource of resources” — staff ask it which internal tool or approach to use for a given engagement. Think of it as an always-available senior who knows where everything is.
Tax: 55% more returns per preparer
On the compliance side, unnamed mid-sized firms using AI tax prep modules reported preparing 55% more returns per preparer without adding staff. The AI auto-maps client QuickBooks data to tax form line items, pulls in scanned documents, and runs error checks. Returns arrive at partner review already clean. One firm reported 90% fewer compliance errors year-over-year.
Nobody got laid off. Staff got reassigned to advisory projects in April instead of grinding through data entry. That’s the real story.
Audit: PKF’s digital worker for investment accounts
PKF O’Connor Davies built a hybrid AI and RPA “digital worker” that logs into over 100 investment accounts, pulls data, consolidates it, reconciles it to the general ledger, flags anomalies, and produces an audit trail report — fully automated. This replaced approximately 500 hours of manual work per year. Audit staff now spend that time on risk assessment instead of data gathering.
This is what agentic AI looks like in audit — not a chatbot answering questions, but an orchestrator executing a defined workflow across multiple systems.
Audit: AI confirmation services
A mid-sized Midwest firm implemented an AI confirmation service that autonomously sends audit confirmations to banks, lawyers, and third parties — and then chases them. The AI handled 90% of follow-up. Audit seniors who used to dread confirmation season were freed to focus on analyzing responses instead. The firm cut one full week off its average audit timeline.
This one matters for morale too. Confirmation chasing is the kind of task that grinds people down. Removing it — not just automating it, but genuinely removing it from someone’s plate — changes how people feel about their work.
Operations: CLA’s two internal AI wins
CliftonLarsonAllen deployed AI on two distinct internal workflows. First: an AI knowledge base that answers staff questions by pulling from past workpapers and manuals, cutting the manager time previously spent on repetitive queries. Second: automated extraction of information from lengthy client PDFs for audits, replacing a staff member who had been spending half their day on copy-paste tasks.
Neither of these is glamorous. Both of them matter.
Two things this changes for the profession
The billing model
When AI cuts the time to complete a return or an audit, hourly billing becomes a problem. A return that used to take 10 hours and billed at $1,000 now takes 5. The client still gets the same value. But if you’re billing hourly, you just cut your revenue in half.
This is why only about 3% of firms now bill purely by the hour for tax prep. Fixed-fee and value-based pricing are the answer — charge for the outcome, not the time. CFOs are watching this closely too — 79% say at least 25% of their accounting workload is already being handled by agentic AI tools.
The staffing pyramid
AI is best at the work that entry-level accountants have traditionally done: data entry, transaction categorization, document extraction, initial reconciliations. That’s not a coincidence — it’s a structural problem.
My first job at PwC involved rolling over prior-year workpapers for fund-of-fund tax work. I didn’t love it. But I learned a ton — how the files worked, how the systems worked, how the data flowed, how the team operated. That foundational experience shaped everything that came after.
When AI automates that work, where do new professionals get those foundational skills? The report projects 15–20% fewer entry-level accounting positions by 2027 at firms fully embracing AI. That’s a structural shift the profession hasn’t fully reckoned with yet.
Key takeaways
- AI adoption in accounting firms jumped from 9% to 41% in one year — the tipping point is already here
- Agentic AI acts — it doesn’t just suggest. It executes workflows autonomously within defined parameters
- RSM, PKF, CLA, and GWCPA have all deployed purpose-built AI for specific high-volume pain points — and the results are concrete
- Hourly billing doesn’t survive AI adoption — fixed-fee and value-based pricing is the direction
- The apprenticeship model for junior accountants is at risk — automating entry-level work removes the foundation of how professionals learn
- The question isn’t whether AI is coming to your firm. It’s whether you’re going to be the one running it
Want the CPE credit? Take the full lesson on EverydayCPE and earn 0.2 CPE credits: [lesson link]
Leave a Reply