AI for Accountants
AI for accountants: what actually works in a CPA firm.
Skip the AI-will-replace-your-job headlines. Here's the boring, useful version: where AI saves real hours in a CPA firm, what it should never touch, and how to roll it out without creating a confidentiality problem.
Where AI pays off first
The unglamorous work is the good work.
None of this is exotic. It's the repetitive drafting and first-pass work that eats a senior's afternoon — handed to a tool that doesn't get tired of it.
Client email & document requests
Drafting the fifth reminder for missing 1099s doesn't need a human brain. AI drafts the email, someone hits send.
Workpaper summaries & memo drafting
First-draft memos and summaries from workpapers and trial balances, so staff start from something instead of a blank page.
Bookkeeping cleanup triage
Flag miscategorized transactions and likely errors in a messy QuickBooks file before a human does the real cleanup.
Engagement letters & proposals
First drafts of engagement letters and proposals from a template and a few details, instead of starting over each time.
Internal knowledge lookup
"What's our firm's position on X?" answered from your own memos and precedent instead of hunting through old emails.
What NOT to hand to AI
Final review stays human. Full stop.
AI drafts. It doesn't sign. Final review of any return, any judgment call on a gray-area position, and anything that actually gets filed needs a reviewer who's accountable for it — not a model that can't be held to a standard.
This isn't caution for its own sake. A model producing a confident, wrong answer is the normal failure mode, not the exception. Treat every output as a draft from a fast but unaccountable junior.
Final review of returns
A licensed reviewer signs off on everything that goes out. AI can draft; it doesn't get the pen.
Judgment calls
Gray-area positions, aggressive vs. conservative treatment — that's a partner decision, not a prompt.
Client data, outside business-tier tools
Only put client data into tools with a business tier and a no-training-data policy in writing. Consumer chatbots don't qualify.
Busy season math
The arithmetic firms actually care about.
Say six seniors each spend 45 minutes a day drafting emails, memos, and first-pass documents that AI could produce in a fraction of the time with a quick human edit. That's 4.5 hours a day across the team — roughly 22 hours a week, every week, from January through April. That's not a hypothetical productivity gain. It's time a senior gets back to do senior work: review, client conversations, the stuff that actually needs a CPA.
The math
6 seniors × 45 min/day drafting
= 4.5 hrs/day recovered
4.5 hrs/day × 5 days
≈ 22 hrs/week recovered
Rollout path
Pilot, then policy, then the whole firm.
Firms that skip straight to "everyone use whatever tool you want" end up with client data in the wrong places and no way to audit what happened. Do it in order.
01
Pilot with 2 staff
Pick one senior and one staff accountant. Give them one narrow use case — email drafts or memo summaries — for a few weeks.
02
Write the firm policy
What tools are approved, what data can and can't go in, and who reviews AI-assisted work before it goes out. In writing, before wider rollout.
03
Train the rest of the firm
Once the policy exists and the pilot has real examples, train everyone else on the same narrow set of use cases — not a free-for-all.
If your firm needs the policy piece before anything else, start with our AI policy template. For hands-on workshops that get your team using these tools correctly, see AI training.
FAQ
Questions CPA firms ask us.
Is this a client confidentiality problem?
It can be, if you use the wrong tool. Consumer-grade chatbots may use your inputs to train their models — never put client data in those. Business-tier tools with a no-training-data policy and a signed data agreement are a different situation. That distinction should be in your firm's written AI policy, not left to individual judgment.
Does this create a professional standards problem?
AI doesn't change your responsibility for what goes out under your firm's name — a human still has to review and sign off on anything filed. Rules on AI use vary by state and by engagement type, so check with your professional liability carrier and your state board before finalizing a policy, not after.
Which tools should we actually use?
Business-tier plans of mainstream AI tools, paired with whatever your firm already runs day to day. The specific product matters less than the policy around it — no-training data terms, a signed agreement, and clear rules on what can and can't be entered.
What does this cost?
Software is usually a modest monthly cost per seat. The bigger cost is staff time to pilot, write the policy, and train the team properly — which is exactly the part firms tend to skip and then regret.
Find out what AI can do for your firm.
Ten minutes, free, no sales pitch. The assessment shows you the highest-leverage AI opportunities in your firm — whether you work with us or not.